Skip to main content

Full text of "BIOLOGÍA"

See other formats

The Foundations of Ethology 

Konrad Z. Lorenz 

The Foundations 
of Ethology 

Translated by Konrad Z. Lorenz 
and Robert Warren Kickert 

With 34 Figures 

Springer Science+Business Media, LLC 

Professor Dr. Konrad Z. Lorenz 
Dr. Robert W. Kickert 
Vienna, Austria 

Frontispiece photo by Hans J. Bóning, Wien 

This English edition is a revised and enlarged versión of Vergleichende 
Verhaltensforschung: Grundlagen der Ethologie, first published in 1978 by Springer- 
Verlag/Wien—New York. 

Library of Congress Cataloging in Publication Data 
Lorenz, Konrad. 

The foundations of ethology. 

Based on a translation of Vergleichende 
Verhaltensforschung, with revisions. 

Bibliography: p. 

Ineludes Índex. 

1. Animáis, Habits and behavior of. 

2. Psychology, Comparative. I. Title. 

QL751.L7213 156 81-5735 


© 1981 by Springer Science+Business Media New York 
Originally published by Springer-Verlag New York Wien in 1981 
Softcover reprint of the hardcover lst edition 1981 

All rights reserved. No part of this book may be translated or reproduced in 
any form without written permission from Springer Science+Business Media, LLC, 

The use of general descriptive ñames, trade ñames, trademarks, etc. in this 
publication, even if the former are not especially identified, is not to be taken 
as a sign that such ñames, as understood by the Trade Marks and Merchandise 
Marks Act, may accordingly be used freely by anyone. Printed in the United 
States of America. 


ISBN 978-3-211-99936-3 ISBN 978-3-7091-3671-3 (eBook) 
DOI 10.1007/978-3-7091-3671-3 

To Nikolaas Tinbergen 


This book is a contribution to the history of ethology—not a definí ti ve 
history, but the personal view of a major figure in that story. It is all the 
more welcome because such a grand theme as ethology calis for a range 
of perspectives. One reason is the overarching scope of the subject. Two 
great questions about life that constitute much of biology are "How does 
it work (structure and function)?" and "How did it get that way (evolu- 
tion and ontogeny)?" Ethology addresses the antecedent of "it." Of what 
are we trying to explain the mechanism and development? Surely behav- 
ior, in all its wealth of detail, variation, causation, and control, is the 
main achievement of animal evolution, the essential consequence of 
animal structure and function, the raison d'etre of all the rest. Ethology 
thus spans between and overlaps with the ever-widening circles of ecol- 
ogy over the eons and the ever-narrowing focus of physiology of the 

Another reason why the history of ethology needs perspectives is the 
recency of its acceptance. For such an obviously major aspect of animal 
biology, it is curious how short a time—less than three decades—has 
seen the excitement of an active field and a substantial fraternity of work- 
ers, the addition of professors and courses to departments and curricula 
in biology (still far from universal), and the normal complement of spe- 
cial journals, symposia, and sessions at congresses. Of course, animal 
behavior has been a subject of serious writings, usually called natural 
history, for centuries. However, for reasons that need historical perspec¬ 
tives to be evaluated, the dignity of a major discipline long escaped all 
but the facets embraced by psychologists. 

Did the arrival and spread of acceptance of the modern study of free 
animal behavior await the ímpetus of a school, of advocates, of theories 



and models? Surely Konrad Lorenz was a major factor, through his 
thought-provoking approach, his eloquence, his inspiration to many stu- 
dents—and above all his intimate and wide familiarity with animáis in 

When a figure so largely responsible for the emergence and develop- 
ment of a scientific approach chooses to summarize the field, as Konrad 
Lorenz does here for ethology, it is an historical document and worth 
attention. It is all the more important for a subject so bound up with 
concepts, conceptual methodology, and interpretation, and so dependent 
on accumulated experience in watching animáis. It is Lorenz's perception 
that "fashion" and "ideological prejudice" have obscured our familiarity 
with some of the basic foundations of ethology, even on the part of 
knowledgeable authors, so that a main aim of this work is to remind us 
of the historical origins and of "how narrow that factual foundation is 
and henee how needful is thorough verification." 

One cannot but wonder whether the brigades of workers who have 
followed Lorenz's lead have not provided either that verification or some 
modification. It is therefore an understandable hope of some readers to 
find Lorenz's evaluation of newer data and his position on the debates 
over concepts or terms. However, this would have meant virtually a full 
review of the accumulated literature and such is not Lorenz's aim within 
these covers. On the whole, apart from a few interesting admissions of 
changes in viewpoint and terminology, one has to look hard for changes 
in usage, definition, or concepts made as a result of the findings and rein- 
terpretations of later workers. 

It is only to be expected that such a document will be highly personal. 
Moreover, it is bound to sound familiar. But a cióse reader of Lorenz will 
note many new or refined positions. Of special interest to those who 
think of ethology as mainly concerned with unlearned, instinctive 
behavior of non-mammalian taxa will be two features. One is a series of 
chapters devoted to adaptive modification of behavior. In one of these 
Lorenz distinguishes between two forms of operant conditioning, a form 
which he believes is common in nature, and another that is quite uncom- 
mon. The other feature is an appendix devoted to Homo sapiens. Here he 
gives reason to claim that the Science of human ethology has much to 
teach us, all the more so because the human species, while not outside 
the natural biological order, is very special indeed in cultural heritage 
and the elaboration of language. 

Only a personal restatement of the foundations by one of the founders 
can have the poignaney this book has as the result of changes in the pre- 
vailing intellectual climate, such as the new respectability of discussions 
about animal consciousness and mental life. Not that such changes have 
an easily predictable effect, for example, in making Lorenz's viewpoint 
more persuasive or some ogres less like straw men. It is certainly one of 
the charms of this book to see for oneself what the arguments sound like 



in the light of all that has happened in Science and society. Lorenz lets 
us reexamine not only the factual foundations but the methodological 
and even some philosophical bases. The reexamination may not always 
lead to greater sympathy for the Lorenzian alternative but should lead to 
better history and a greater motivation to broaden the base. 

Appreciation of this book and the wealth of observation of nature it 
represents is enhanced if one adapts to Lorenz's methods of exposition. 
Each reader will no doubt develop his or her own approach. For example, 
I learned not to judge or even to try to absorb the sweeping induction 
that opens the paragraph, but to wait until the example unfolds that 
makes the key terms understandable. It is best not to be put off by expres- 
sions such as "explanatory monism," "atomism," "technomorphically," 
"teleonomy," or "relatively entirety-independent." But these are no 
doubt the limitations of a relatively example-dependent physiologist- 
reader! I could not expect every familiar term to be defined or strong 
language to be eschewed ("never," "any at all," "exploded"). And I 
learned that one road to understanding is to contrast the right view with 
a delineation of the bad guy's view. 

There have been various images of scientists: bricklayers adding facts 
to erect an edifice of knowledge, revolutionaries putting together dis- 
crepancies to overturn established "paradigms," delvers uncovering nug- 
gets of gold, and wise men avoiding the Baconian idols of the market- 
place and idols of the tribe. This book conjures up the image of a slayer 
of dragons; the preferred form of exposition of scientific advance is to 
show how wrong was some previous view. Most authors cited are either 
protagonists or adversaries, and the latter are astonishingly bad. The list 
of bad guys is long: vitalists, monists, reflexologists, Pavlovists. 

One reads on, captivated by the parade of striking examples. The 
enduring images are the vivid descriptions—of a shrike holding an 
object in its bilí and searching for thorns, or performing impaling 
motions in vacuo; of a goose, satiated with corn but deprived of an 
opportunity to up-end and gather food from the pond bottom, so that 
when offered corn thrown into the pond at the right depth, began "feed- 
ing for the sake of up-ending instead of up-ending for the sake of feed- 
ing"; of the experienced male cichlid fish that is not fooled by the best 
fish-like dummy, in contrast to the male raised in isolation that gives a 
full male courtship response when the simplest dummy moves into view. 
While unusual for fish, this drives home both the nature of the evidence 
for innate releasing mechanisms and the possibility of refining both the 
sign stimuli and the recognition process by learning. These and scores of 
other graphic instances are mostly familiar to ethologists as classical 
examples; they form the solid empirical foundations of ethology. It is 
understandable how they invite categorization, concept formation, the 
induction of entities, and the attempt to interpret in some sort of crude 
mechanistic terms. It is too much to ask that the ethologist "stick to his 



last" and avoid wandering into physiology. We should expect and wel- 
come such excursions, and judge their success not so much by the sophis- 
tication or modernity of the physiological model but by how heuristic it 
is in suggesting things to look for. 

Rarely does a founder of a field give us his insider's view of it. Would 
that other participants in the founding of ethology were moved to 
emulate Dr. Lorenz in sharing their impressions of its development. In 
such a spectrum of memoirs, this book would have a key position. 

Theodore Holmes Bullock 
University of California, San Diego 
La Jolla, California 


In some respects, the development of a Science resembles that of a coral 
colony. The more it thrives and the faster it grows, the quicker its first 
beginnings—the vestiges of the founders and the contributions of the 
early discoverers—become overgrown and obscured by their own prog- 
eny. There is one drawback to the strategy of growth pursued by the 
coral tree. The polyps at the end of its branches have a much better 
chance of further development than those situated near the foundation. 
The ends go on growing faster and faster without considering the neces- 
sity for strengthening, in proportion, the base that must carry the weight 
of the whole structure. Unlike an oak tree, the coral colony does nothing 
to solidify its support. Consequently, there is a lot of coral rubble 
detached from points of departure, and this is either dead or, if still 
partly alive, growing in indeterminate directions and getting nowhere. 

Having myself grown very near the point from which ethology, as a 
new branch of biological Science, had its origin, it may seem presump- 
tuous if I compare the present State of our Science to a coral colony whose 
branches, by losing contact with their foundation, are producing quite a 
lot of rubble. However, there is no doubt that they do, and I am pre- 
sumptuous enough to criticize this. My justificaron lies in the fact that 
really important discoveries, such as those made by Charles Otis Whit- 
man, Oskar Heinroth, Erich von Holst, Kenneth Roeder, and others, are 
being forgotten, and for reasons, I contend, which are partly to be found 
in mere fashion, and partly in ideological prejudices. 

So this book is not an up-to-date textbook on ethology. It does not pre¬ 
sume to inelude all the most recent developments within this Science, 
not even those which mean very real advances. Its aim is to remind sci- 
entists in general of the basic faets on which ethology is founded, and to 
remind ethologists in particular of how narrow the factual foundation of 
our Science really is, how completely our Science relies on these faets 
being correct, and how much, therefore, their thorough verification 
remains necessary. 


Foreword vii 

Preface xi 

Introductory History 1 

Part One Methodology 13 

Chapter I Thinking in Biological Terms 15 

1. The Diíferences Between the Goals of Physical and Biological 

Research 15 

2. The Limits of Reduction 17 

3. Ontological Reductionism 19 

4. The Evolutionary Event as a Limitation of Reduction 22 

5. The Question "What For?" 23 

6. Teleological and Causal Views of Nature 32 

Chapter II The Methodology of Biology and Particularly of 

Ethology 36 

1. The Concept of a System or an Entirety 36 

2. The Sequence of Cognitive Steps Dictated by the Character of 

Systems 38 

3. The Cognitive Capacity of Perception 40 

4. So-Called Amateurism 46 

5. Observing Animáis in the Wild and in Captivity 47 

6. Observing Tame Animáis Not Kept Captive 51 

7. Knowing Animáis: A Methodological Sine Qua Non 52 



8. The Non-Obtrusive Experiment 53 

9. The Deprivation Experiment 57 

10. The Relatively Entirety-Independent Component 64 

Chapter III The Fallacies of Non-System-Oriented Methods 66 

1. Atomism 66 

2. Explanatory Monism 67 

3. Operationalism and Explanatory Monism of the Behaviorist School 68 

Chapter IV The Comparative Method 72 

1. Reconstruction of Genealogies 72 

2. Criteria of Taxa 74 

3. The Hypothesis of Growth 81 

4. Documentation Through Fossils 81 

5. Homology and Its Criteria 85 

6. The Number of Characteristics as a Criterion of Homology 87 

7. Convergent Adaptation 88 

8. Analogy as a Source of Knowledge 89 

9. Homoiology 93 

10. Systematics and the Need for Great Numbers of Characteristics 93 

11. The Changing Valué of Single Characteristics 96 

12. The Difficulties and the Importance of "Microsystematics" 98 

13. The Origin of Ethology 100 

14. Chapter Summary 101 

Part Two Genetically Programmed Behavior 105 

Chapter I The Centrally Coordinated Movement or Fixed 

Motor Pattern 107 

1. History of the Concept 107 

2. Differences in Intensity 110 

3. Qualitatively Identical Excitation Activating Different Motor 

Patterns 112 

4. Unity of Motivation 113 

5. The Method of Dual Quantification 115 

6. Action-Specific Fatigue 118 

7. Threshold Lowering of Releasing Stimuli 123 

8. Effects Obscuring the Accumulation of Action-Specific Excitability 125 

9. Vacuum Activity 127 

10. Appetitive Behavior 129 

11. Threshold Lowering and Appetitive Behavior in Avoidance 130 

12. Driving and Being Driven 133 

13. Neurophysiology of Spontaneity 136 

14. Analogies of Function in Neural Elements and Integrated Systems 145 

15. Chapter Summary 148 

Contents xv 

Chapter II Afferent Processes 153 

1. The Innate Releasing Mechanism (IRM) 153 

2. Limits to the Functions of IRMs 162 

3. IRM and the Releaser 166 

4. An Important Rule of Thumb 170 

5. IRMs Rendered More Selective by Learning 173 

Chapter III The Problem of the "Stimulus" 176 

1. All-Embracing Conceptualizations 176 

2. Stable and Spontaneously Active Nervous Elements 176 

3. Analogous Phenomena in Integrated Neural Systems 179 

4. Action-Specific Potential (ASP) 184 

Chapter IV The Behavior Mechanisms Already Described Built 

into Complex Systems 189 

1. Appetitive Behavior Directed at Quiescence 189 

2. Searching Automatism 191 

3. Hierarchical Systems 193 

4. The Relative Hierarchy of Moods 202 

5. The Locus of "Superior Command" (Übergeordnete Kommandostelle) 204 

Chapter V How Unitary Is "An Instinct"? 211 

1. The Danger of Naming Instincts by Their Functions 211 

2. The Multiplicity of Motivations 212 

3. Integrating Effect of the Instinct Hierarchy 215 

4. Interaction Between Motor Patterns 216 

5. Motor Patterns Not Specific to the System 217 

6. Chapter Summary 219 

Chapter VI Mechanisms Exploiting Instant Information 221 

1. Receiving Information Does Not Always Mean Learning 221 

2. The Regulating Cycle or Homeostasis 222 

3. Excitability 223 

4. Amoeboid Response 223 

5. Kinesis 225 

6. Phobic Response 225 

7. Topical Response or Taxis 227 

8. Telotaxis or "Fixating" 229 

9. Temporal Orientation 231 

10. Navigation by Sextant and Chronometer 233 

11. Taxis and the Fixed Motor Pattern 235 

12. Taxis and Insight 237 

Chapter VII Múltiple Motivation in Behavior 242 

1. The Rarity of Unmixed Motivation 242 

2. Superposition 243 

3. Mutual Inhibition and Alternation 245 

4. Displacement Activities 249 



Part Three Adaptive Modification of Behavior 255 

Chapter I Modification 257 

1. Modification and Adaptive Modification 257 

2. Analogous Processes in Embryogenesis 258 

3. Learning as an Adaptive Modification 259 

Chapter II Learning Without Association 263 

1. Facilitation and Sensitization 263 

2. Habituation or Stimulus Adaptation 265 

Chapter III Learning Through Association Without Feedback 

Reporting Success 268 

1. Association 268 

2. Habituation Linked with Association 269 

3. "Becoming Accustomed" or Habit Formation 272 

4. The Conditioned Reflex Proper or Conditioning with Stimulus 

Selection 276 

5. Avoidance Responses Acquired Through Trauma 278 

6. Imprinting 279 

7. Conditioned Inhibition 284 

8. Chapter Summary 286 

Chapter IV Learning Eífected by the Consequences of Behavior 289 

1. The New Feedback 289 

2. Minimum Complication of the System 293 

3. Conditioned Appetitive Behavior 295 

4. Conditioned Aversión 300 

5. Conditioned Action 303 

6. Conditioned Appetitive Behavior Directed at Quiescence 306 

7. Operant Conditioning (In the Sense Here Advocated) 309 

8. Chapter Summary 312 

Chapter V Motor Learning, Voluntary Movement, and Insight 315 

1. Motor Learning 315 

2. So-Called Voluntary Movement 319 

3. Voluntary Movement and Insight 323 

Chapter VI Exploratory Behavior or Curiosity 325 

1. Choice of Behavior Patterns 325 

2. The Autonomous Motivation of Exploratory Behavior 326 

3. Latent Knowledge 327 

4. Objectivity 328 

5. Specialization for Versatility 328 

6. Play 329 

7. Curiosity, Play, Research, and Art 333 



Afterword to Part Three 336 

Appendix Concerning Homo sapiens 338 

1. Anthropologists' Allegations 338 

2. On Analogies 338 

3. The Difference of Homo sapiens 339 

4. Conceptual Thought and Syntactic Language 342 

5. Consequences 343 

6. Cultural Ethology 344 

References 347 

Index 363 

Introductory History 

Ethology, the comparative study of behavior, is easy to define: it is the 
discipline which applies to the behavior of animáis and humans all those 
questions asked and those methodologies used as a matter of course in 
all the other branches of biology since Charles Darwin's time. 

When one considers with what rapidity the ideas of evolution, and 
particularly the Darwinian concept of natural selection, caught on in 
almost all branches of biology, one searches for an explanation as to why 
these ideas were so tardily accepted by the disciplines of psychology and 
behavioral Science. The main reason that biological thinking and espe- 
cially comparative methods were prevented from penetrating the study 
of behavior was an ideological dispute between two prominent schools 
of psychology. 

The bitterness with which this dispute was pursued was nourished, 
above all, by the diverse philosophies of the antagonists. The school of 
purposive psychology, represented primarily by William MacDougall 
and later by Edward Chase Tolman, postulated an extranatural factor: 
"instinct" was regarded as an agens or agency neither in need of ñor 
accessible to a natural explanation. "We consider an instinct but we do 
not explain it," wrote Bierens de Haan as late as 1940. To this conception 
of instinct was always also appended a belief in its infallibility. Mac¬ 
Dougall rejected all mechanistic explanations of behavior. For example, 
he considered it a consequence of instinct when insects pressed forward 
purposively toward light; he conceded the possibility of a mechanistic 
explanation, through tropism, only in those cases where these animáis, 
most unpurposively, flew into a burning lamp. According to MacDougall 
and his school, everything animáis do is in pursuit of a purpose and this 
purpose is set by their extranatural and infallible instinct. 


Introductory History 

Those of the behaviorist school of psychology justifiably criticized the 
assumption of extranatural factors as unscientific. They demanded causal 
explanations. Through their methodology they sought to place them- 
selves as much apart as possible from the purposive psychologists. They 
regarded the controlled experiment as the only legitímate source of 
knowledge. Empirical methods were to take the place of philosophical 

With the exception of a certain lack of appreciation for simple obser¬ 
varon, this program incorporated no methodological error, and yet it 
brought about an unfortunate consequence: all research interests were 
concentrated on those aspects of animal and human behavior which 
readily lent themselves to experimentaron—and this led to explanatory 

A combination of William Wundt's (1922) association theory with the 
reflex theory (reflexology) that was then dominating the fields of physi- 
ology and psychology, as well as with the findings of I. P. Pavlov (1927), 
facilitated the abstraction of a behavior mechanism—the so-called con- 
ditioned reflex —the qualities of which marked it as ideal for experimental 

At that time the corrective criticism made by the behaviorists concern- 
ing the opinions held by the purposive psychologists was salutary in 
every way. But, unobserved, a ruinous logical error crept into behaviorist 
thinking: because only learning processes could be examined experimen- 
tally and since all behavior must be examined experimentally, then, con- 
cluded those of the behaviorist school, all behavior must be learned— 
which, naturally, is not only logically false but also, factually, complete 

Knowing the views of those in the opposition, and having made a jus- 
tifiable critique of those views, the purposive psychologists as well as the 
behaviorists were pushed into extreme positions which neither of them 
would otherwise have taken. While those of one group were imbued 
with a mystical veneration of "THE instinct" and attributed excessive 
capacities, even infallibility, to the inborn, those of the other group 
denied its very existence. The purposive psychologists, who were quite 
aware of innate behavior patterns, regarded everything instinctive as 
inexplicable and, just as Bierens de Haan (1940), refused even to attempt 
a causal analysis. Those others who certainly would have been capable 
and ready to undertake such analytical research denied the existence of 
anything inborn, and dogmatically declared that all behavior was 
learned. The truly tragic aspect of this historical situation is that the pur¬ 
posive psychologists, particularly MacDougall himself, knew animáis 
well and possessed a good, general knowledge of animal behavior, some- 
thing which is still lacking among the behaviorists even today, because 
they regard simple observation as unnecessary, in fact, as "unscientific." 
In this context, the truth of a statement of Faust's comes to mind: "What 

Introductory History 


one does not know is exactly what one needs, and what one does know 
one cannot use." 

This ideological dispute between these two schools of psychology was 
still being actively pursued when, completely unnoticed by both and 
independent of their influence, the scientific study of innate behavior 
patterns carne into being. At the turn of the century Charles Otis Whit- 
man and, a few years later, independently of him, Oskar Heinroth dis- 
covered the existence of patterns of movement, the similarities and diífer- 
ences of which, from species to species, from genus to genus, even from 
one large taxonomic group to another are retained with just as much con- 
stancy and in exactly the same way as comparable physical characters. In 
other words, these patterns of movement are just as reliably characteristic 
of a particular group as are tooth and feather formation and such other 
proven distinguishing physical attributes used in comparative morphol- 
ogy. For this fact there can be no other explanation than that the simi¬ 
larities and dissimilarities of these coordinated movements are to be 
traced back to a common origin in an ancestral form which also already 
had, as its very own, these same movements in a primeval form. In short, 
the concept of homology can be applied to them. 

These facts alone prove that these movements originate phylogeneti- 
cally and are imbedded in the genome. It is just this that is overlooked 
by those students of behavior who would like to explain away every con¬ 
ceptual distinction between innate and acquired characteristics. When 
the African black duck (Anas sparsa) living on tropical rivers, the mallard 
living on our own lakes, the many species of wild ducks living on the 
ponds of zoos, and the domesticated ducks living in the barnyards of our 
farms, in spite of the diíferences of their environments and despite all 
the possible influences of captivity, display courtship movements that 
are unmistakably similar in a countless number of characteristics, then 
the program for these movements must be anchored in the genome in a 
manner exactly identical with that in which the program of morpholog- 
ical characters is coded in the genes. If, after this discovery, theories con¬ 
cerned with the problem of "nature versus nurture" continué to be pub- 
lished, this is explainable only through the assumption that these 
authors are unaware of the discoveries made at the turn of the century, 
or that they have chosen to ignore them. That they indeed do this was 
soon clear to me. American psychologists often visited Karl Bühler's 
institute. I asked each and every one of them whether they knew the 
ñame Charles Otis Whitman. Not one of them knew the ñame. 

The discovery that movement patterns are homologous is the Archi- 
medean point from which ethology or the comparative study of behavior 
marks its origin. Paradoxically, even the work of authors who deny the 
essential diíference between innate and acquired behavior mechanisms 
is built upon the same factual base. 

I discovered for myself, independently of Whitman and Heinroth, that 


Introductory History 

patterns of movement are homologous. When studying at the university 
under the Viennese anatomist, Ferdinand Hochstetter, and after I had 
become thoroughly conversant with the methodology and procedure of 
phylogenetic comparison, it became immediately clear that the methods 
employed in comparative morphology were just as applicable to the 
behavior of the many species of fish and birds I knew so thoroughly, 
thanks to the early onset of my love for animáis. Soon after this I met 
Oskar Heinroth, the discoverer of my discovery, and early in the 1930's 
both of us learned through communication with the American ornithol- 
ogist, Margret Morse Nice, that Charles Otis Whitman had come to essen- 
tially the same conclusions as Heinroth about ten years earlier. At the 
same time all this was happening, we met the most distinguished of 
Whitman's students, Wallace Craig. 

Neither Whitman ñor Heinroth ever expressed any views concerning 
the physiological nature of the homologous movement patterns they had 
discovered. My own knowledge of the physiology of the central nervous 
system carne from lectures and textbooks in which the Sherringtonian 
reflex theory (1906) was regarded as the last word and the incontestable 
truth. The expression "reflex" evokes, linguistic-logically, the visión 
of a simple, linear causal relationship between the incoming stimulus 
and the response given to it by the organism. In this simplicity lies the 
seductive eífect of the concept: It is just as easy to understand as it is 
to teach. 

Under Karl Bühler's tuition I gained enough knowledge of the two 
prominent schools of American psychology to feel myself qualified to 
criticize them on two fundamental points. The first was that the infalli¬ 
ble, preternatural "instinct" in which the purposive psychologists 
believed simply did not exist; too often had I seen innate behavior pat¬ 
terns taking their course in completely blind and senseless sequences. 
The second criticism was that the point of view of the behaviorists, that 
all animal behavior is learned, was totally false. 

I had published several short articles, based on my own observations, 
about the problem of the innate and homologous in motor patterns when 
my friend, Gustav Kramer, imposed himself on the course of these events 
by influencing the biologist, Max Hartmann, to invite me to give a lee- 
ture to the Kaiser Wilhelm Society for the Advancement of Science (now 
the Max Planck Society). Kramer was carrying out his intention of pro- 
viding a setting for a discussion between Erich von Holst and me. He 
was von Holst's friend as well as mine, and he was well aware that the 
phenomena which I was observing in the motor patterns of intact ani¬ 
máis were very closely related to those processes which von Holst was 
investigating experimentally at the neurophysiologic level. Gustav Kra¬ 
mer believed that the congruity between von Holst's research results and 
mine would be that much more startling and convincing the longer we 

Introductory History 


worked completely independently of one another; that is why he per- 
petuated this remarkable feat of extended reticence. 

So then, in 1935, I gave my lecture at Harnack House in Berlín. Its 
theme was based on my article, "The Concept of Instinct Then and 
Now." (1932) There I made it clear that any animal is perfectly capable, 
through goal oriented and variable behavior, of striving toward a pur- 
pose, but that this purpose may not, as the purposive psychologists sup- 
posed, be equated with the achievement of the teleonomic function of 
behavior. The purpose toward which the animal, as subject, is striving, 
is simply a run-through or discharge of that kind of innate behavior 
which Wallace Craig designated as "consummatory action" (1918) and 
which we now cali the drive-reducing consummatory act. Up to this 
point what I said then is more or less what I believe today. 

But what I had to say about the physiological nature of fixed action 
patterns was influenced by doctrinaire bias. Led by MacDougall, the pur¬ 
posive psychologists had continued their battle against the reflex theory 
of the behaviorists and, quite rightly, had emphasized the spontaneity of 
animal behavior. "The healthy animal is up and doing," MacDougall had 
written. I was already thoroughly familiar with the writings of Wallace 
Craig and, through my own research, I was well acquainted with the 
phenomena of appetitive behavior and of threshold lowering for releas¬ 
ing stimuli—and I should have borne in mind a particular sentence of a 
letter Craig had sent shortly before, in which he had argued against the 
reflex concept: "It is obviously nonsense to speak of a re-action to a stim- 
ulus not yet received." 

At that juncture mere common sense ought to have prompted me to 
put the following question: Innate motor patterns have, apparently, 
nothing to do with higher intellectual capacities; they are góverned by 
central nervous processes which occur quite independently of external 
stimuli and they tend to be repeated rhythmically. Do we know of any 
other physiological processes which function in a similar way? The 
obvious answer would have been: Such motor patterns are very well 
known, particularly those of the vertébrate heart for which stimulus pro- 
ducing organs are anatomically known and the physiology of which has 
been thoroughly studied. 

I lacked the independence of mind and the self-assurance that would 
have been necessary to ask this question. My valid aversión toward the 
preternatural and inexplicable factors which the vitalists had summoned 
to interpret spontaneous behavior was so deep that I lapsed into the 
opposite error; I assumed that it would be a concession to the vitalistic 
purposive psychologists if I were to deviate from the conventional 
mechanistic concept of reflexes, and this concession I did not wish to 
make. During the course of that lecture I did cover completely, and with 
especial emphasis, all those characteristics and capacities of fixed action 


Introductory History 

patterns which could not be accounted for by means of the chain reaction 
theory, yet, in my summary at the end, I still concluded that fixed action 
patterns depended on the linkage of unconditioned reflexes even if the 
cited phenomena of appetitive behavior, threshold lowering, and vac- 
uum activity (I will return to this on page 127) would require a supple- 
mentary hypothesis for clarification. 

Sitting next to my wife in the last row of the auditorium was a young 
man who followed the lecture intently and who, during the exposition 
on spontaneity, kept muttering, "Menschenskind! That's right, that's 
right!" However, when I carne to the concluding remarks described 
above, he covered his head and groaned, "Idiot." This man was Erich 
von Holst. After the lecture we were introduced to one another in the 
Harnack House restaurant and there it took him all of ten minutes to 
convince me forever that the reflex theory was indeed idiotic. 

The moment one assumed that the processes of endogenous produc- 
tion and central nervous coordination of impulses, discovered by von 
Holst, and not some linkage of reflexes, constitute the physiological bases 
of behavior patterns, all the phenomena that could not be fitted into the 
reflex theory, such as threshold lowering and vacuum activities, not only 
obtained an obvious explanation but became effects to be postulated on 
the basis of the new theory. 

A consequence of this new physiological theory of the fixed motor pat- 
tern was the necessity to analyze further that particular behavioral Sys¬ 
tem which Heinroth and I had called the arteigene Triebhandlung (literally, 
species-characteristic drive-action) and which we had regarded as an 
elementary unit of behavior. Obviously, the mechanism which selec- 
tively responded to a certain stimulus situation must be physiologically 
different from the fixed motor pattern released. As long as the whole 
system was regarded as a chain of reflexes, there was no reason for con- 
ceptually separating, from the rest of the chain, the first link that set it 
going. But once one had recognized that the movement patterns resulted 
from impulses endogenously produced and centrally coordinated and 
that, as long as they were not needed, they had to be held in check by 
some superordinated factor, the physiological apparatus which triggered 
their release emerged as a mechanism sui generis. These mechanisms 
that responded selectively to stimuli, in a certain sense served as "filters" 
of afference, were clearly fundamentally different from those which pro¬ 
duced impulses and from the central coordination that was independent 
of all afference. 

This dismantling of the concept of the arteigene Triebhandlung into its 
component parts signified a substantial step in the development of 
ethology. The step was taken in Leyden at a congress called together by 
Prof. van der Klaauw. During discussions that lasted through the nights, 
Niko Tinbergen and I conceived the concept of the innate releasing mech¬ 
anism (IRM), although it is no longer possible to determine by which one 

Introductory History 


of us it was born. Its further elaboration and refinement, and the explo¬ 
raron of its physiological characteristics, especially its functional limi- 
tations, are all due to Niko Tinbergen's experiments. 

Concurrent with the conceptualization of the IRM, the concept of the 
fixed action pattern or instinctive motor pattern was also narrowed and 
made more precise, and in exactly the way Charlotte Kogon had pro- 
posed as early as 1941 in her book, The Instinctive as a Philosophical Problem, 
a book which regrettably remained unknown to me until just recently. 
Subsequently, and up to the present, the concepts of IRM and of the fixed 
motor pattern have proved their worth; their utility in the most diverse 
kinds of flow diagrams make it probable that they are, in fact, function- 
ally if not also physiologically identical mechanisms. For the visualiza¬ 
ron and presentation of hierarchically organized instincts (Tinbergen, 
Baerends, Leyhausen) they have been especially useful. 

During the years that followed ethology developed quickly, both in 
the results achieved and in the increasing number of researchers. A large 
store of data was laboriously assembled; many unique discoveries were 
made. If one chooses to criticize this period of felicitous research, it can 
be reproached for one-sidedness, even for a certain failure to think in 
terms of systems. This was inherent in an orientation that almost com- 
pletely ignored learning processes ; above all, the relationships and inter- 
relationships that existed between the newly discovered inborn behavior 
mechanisms and the various forms of learning were barely touched. My 
modest contribution, which comprised a formulation of the "instinct- 
learning intercalaron" concept, got no further than formulation; besides, 
the example on which the conceptualization—correct in itself—was 
based, was false. (See page 60.) 

In 1953 a critical study appeared which had a behaviorist point of view 
but which did not come from a behaviorist. In "A Critique of Konrad 
Lorenz's Theory of Instinctive Behavior" Daniel S. Lehrmann dismissed, 
on principie, the existence of innate movement patterns and, in so doing, 
supported his argument substantially by using a thesis of D. O. Hebb 
(1940) who had maintained that innate behavior is defined only through 
the exclusión of what is learned and, thus, as a concept was "nonvalid," 
that is, unusable. Drawing on the findings of Z. Y. Kuo (1932), Lehrmann 
also asserted that one could never know whether or not particular behav¬ 
ior patterns had been learned within the egg or in útero. Kuo had already 
recommended abandoning the conceptual separation of the innate and 
the acquired. All behavior, in his opinión, consisted of reactions to stim- 
uli and these reflected the interaction between the organism and its 
environment. The theory of a pre-extant relationship between the organ¬ 
ism and the conditions of its environment is no less questionable, for 
Kuo, than the assumption of innate ideas. 

My answer to Lehrmann's critique was short and forceful but, at first, 
missed the most essential mark. The assertion that the innate in compar- 


Introductory History 

ative studies of behavior is defined only through the exclusión of learn- 
ing processes is entirely false: like morphological traits, innate behavior 
patterns are recognizable through the same systematic distribution of 
attributes; the concepts of innate and acquired are as well defined as 
genotype and phenotype. The reply to the theory that the bird within 
the egg or the mammal embryo within the uterus could there have 
learned behavior patterns which then "fit" its intended environment 
was formulated by my wife and with a single phrase: "Indoor ski 
course." I myself wrote at the time that Lehrmann, in order to get around 
the concept of innate behavior patterns, was actually postulating the 
existence of an innate schoolmarm. 

My formulation of the concept of the "innate schoolmarm" was clearly 
intended as a reductio ad absurdum. What neither I ñor my critics saw 
was that in just this teaching mechanism the real problem was lurking. 
It took me nearly ten years to think through to where, actually, the error 
of the criticism and the counter-criticism was located. It was so very dif- 
ficult to find because the error had been committed in exactly the same 
way by both the extreme behaviorists and by the older ethologists. It was, 
as a matter of fact, incorrect to formúlate the concepts of the innate and 
the acquired as disjunctive opposites; however, the mutuality and inter- 
section of their conceptual contents were not to be found, as the "instinct 
opponents" supposed, in everything apparently innate being, really, 
learned, but the very reverse, in that everything learned must have as its 
foundation a phylogenetically provided program if, as they actually are, 
appropriate species-preserving behavior patterns were to be produced. 
Not only Oskar Heinroth and I, too, but other older ethologists as well, 
had never given much concentrated thought to those phenomena which 
we quite summarily identified as learned or as determined through 
insight and then simply shoved them to the side. We regarded them—if 
one wishes to describe our research methods somewhat uncharitably— 
as the ragbag for everything that lay outside our analytical interests. 

So it happened that neither one of the older ethologists ñor one of the 
"instinct opponents" posed the pertinent question about how it was pos- 
sible that, whenever the organism modified its behavior through learn- 
ing processes, the right process was learned, in other words, an adaptive 
improvement of its behavioral mechanisms was achieved. This omission 
seemed particularly crass on the part of Z. Y. Kuo (1932) who had so 
expressly disassociated himself from every predetermined connection 
between organism and environment but who, at the same time, regarded 
it as axiomatic that all learning processes induced meaningful species- 
preserving modifications. As far as my knowledge goes, P. K. Anokhin 
(1961) was first among the theorists of learning to grasp the conditioned 
reflex as a feedback Circuit in which it was not only the stimulus configu- 
ration arriving from the outside, but more especially the return notifica- 

Introductory History 


tion reporting on the completion and the consequences of the condi- 
tioned behavior that provided an audit of its adaptiveness. 

As in many other cases of erroneous reasoning, the behaviorists' exclu¬ 
sión of questions concerning the adaptive valué of learned behavior may 
be traced to their emphatic antagonism to the school of purposive psy- 
chology. The latter's uninhibited commitment to behavior's extranatural 
purpose created in the behaviorists such antipathy to all concepts of pur- 
pose that, along with purposive teleology, they also resolutely refused to 
consider any species-preserving purposefulness, including teleonomy as 
defined by C. Pittendrigh (1958). This attitude, unfortunately, made 
them blind to all those things that could be understood only through a 
comprehension of evolutionary processes. 

The innate schoolmarm, which tells the organism whether its behavior 
is useful for or detrimental to species continuation and, in the first 
instance reinforces and in the second extinguishes that behavior, must 
be located in a feedback apparatus that reports success or failure to the 
mechanisms of the first phases of antecedent behavior. This realization 
carne to me only slowly and independently of P. K. Anokhin. I published 
my theories on this subject also in 1961 in my monograph, Phylogenetische 
Anpassung und adaptive Modifikation des Verhaltens, which I later extended 
and enlarged for a book in English, Evolution and Modification of Behavior. 
As I emphasized in that publication, whenever a modification of an 
organ, as well as of a behavior pattern, proves to be adaptive to a partic¬ 
ular environmental circumstance, this also proves incontrovertibly that 
information about this circumstance must have been "fed into" the organism. 
There are only two ways this can happen. The first is in the course of 
phylogenesis through mutation and/or new combinations of genetic fac- 
tors and through natural selection. The second is through individual 
acquisition of information by the organism in the course of its ontogeny. 
"Innate" and "learned" are not each defined through an exclusión of the 
other but through the way of entrance taken by the pertinent information that 
is a prerequisite for every adaptive change. 

The bipartition, the "dichotomy" of behavior into the innate and the 
learned is misleading in two ways, but not in the sense maintained in 
the behaviorist argument. Neither through observation ñor through 
experimentation has it been found to be even in the least probable, still 
less a logical necessity, that every phylogenetically programmed behav¬ 
ior mechanism must be adaptively modifiable through learning. Quite 
the contrary, it is as much a fact of experience as it is logical to postúlate 
that certain behavior elements, and exactly those that serve as the built- 
in "schoolmarm" and conduct the learning processes along the correct 
route, are never modifiable through learning. 

But, on the other hand, every "learned behavior" does contain phy¬ 
logenetically acquired information to the extent that the basis of the 


Introductory History 

teaching function of every "schoolmarm" is a physiological apparatus 
that evolved under the pressure of selection. Whoever denies this must 
assume a prestabilized harmony between the environment and the 
organism to explain the fact that learning—apart from some instructive 
failures—always reinforces teleonomic behavior and extinguishes 
unsuitable behavior. Whoever makes himself blind to the facts of evo- 
lution arrives inevitably at this assumption of a prestabilized harmony, 
as have the cited behaviorists and that great vitalist, Jakob von Uexküll. 

The search for the source of information which underlies both innate 
and acquired adaptation has, since those earlier years, yielded significant 
results. I will mention only the research done by Jürgen Nicolai (1970) 
with whydah birds (Viduinae) in which the information can be "coded" 
in such an intricate way: essential parts of the adult bird's song have been 
learned by monitoring the begging tones and other tonal expressions of 
whichever species of host bird the whydah happened to be hatched and 

Inquiry into the phylogenetic programming of thé acquiring processes 
has proved to be important in many respects. Like imprinting, some 
acquiring processes are impressionable only during specific sensitive 
periods of ontogeny; a failure to perceive and meet their needs during 
those crucial periods in animáis and humans can result in irremediable 
damage. Within cultural contexts the distinction between the innate and 
the acquired is also significant. Man, too, and his behavior are not unlim- 
itedly modifiable through learning and, thus, many inborn programs 
constitute human rights. 

As early as 1916, Oskar Heinroth wrote in the conclusión of his classic 
paper on waterfowl: 

I have, in this paper, drawn attention to the behavior used in social inter- 
course and this, especially in birds living in social communities, turns out to 
be quite amazingly similar to that of human beings, particularly in species 
in which the family—father, mother and children—remain together living 
in a cióse unión as long as, for instance, geese do. The taxon of Suropsidae 
[the branch formed by reptiles and birds; see Figure 6, page 75] has here 
evolved emotions, habits and motivations very similar to those which we 
are wont to regard, in ourselves, as morally commendable as well as con- 
trolled by reason. The study of the ethology of higher animáis (still a regret- 
tably neglected field) will forcé us more and more to acknowledge that our 
behavior towards our families and towards strangers, in our courtship and 
the like, represents purely innate and much more primitive processes than 
we commonly tend to assume. 

This early admonishment notwithstanding, ethology was curiously tardy 
in approaching Man as a subject. 

In the investigation of humans it is not easy to fulfill the primary task 
of ethology, which is the analytical distinction of fixed motor patterns. 

Introductory History 


No less a man than Charles Darwin in his monograph, The Expression of 
Emotion in Man and Animáis (1872), pointed out the homology of some 
human and animal motor patterns. The homology was convincing, but 
solid proof still remained necessary. 

Irenáus Eibl-Eibesfeldt (1973) was the first to afford this proof. He 
chose the same movements which Darwin had studied—those express- 
ing emotions. For obvious reasons, the experiments involving social iso- 
lation that are generally used to prove a motor pattern to be independent 
of learning could not be used with humans, so Eibl fell back on the study 
of those unfortunates with whom an illness had already initiated this 
experiment in an equally cruel and effective manner: he studied children 
born deaf and blind. As he was able to demónstrate by means of film 
analyses, these children possessed a practically unchanged repertoire of 
facial expressions although, living in permanent and absolute darkness 
and silence, they had never seen or heard these expressed by any fellow 

As a second route of approach, Eibl-Eibesfeldt (1967, 1968) used the 
cross-cultural method to study the expression of emotions in humans. He 
observed and filmed representatives of as many cultures as he could, in 
standardized situations such as greeting or taking leave, quarreling, 
experiencing grief and enjoyment, courting, and so on. The essential pat¬ 
terns of expressing emotions proved to be identical in all the cultures he 
was able to study, even when the patterns were subjected to minute anal- 
ysis by means of slow motion films. What varied was only the control 
exerted by tradition: this affected a purely quantitative differentiation of 

The most important result of Eibl-Eibesfeldt's extensive and patient 
research can be stated in a single sentence. The motor patterns shown 
undiminished by deaf-and-blind children are identical to those that, 
through cross-cultural investigation, have been shown to be inaccessible 
to cultural change. In view of these incontrovertible results, it is a true 
scientific scandal when many authors still maintain that all human 
expression is culturally determined. 

A strong support for human ethology has come from the unexpected 
area of linguistic studies; Noam Chomsky and his school have demon- 
strated that the structure of logical thought—which is identical to that of 
syntactic language—is anchored in a genetic program. The child does 
not learn to talk; the child learns only the vocabulary of the particular 
language of the cultural tradition into which it happens to be born. 

A surprising and important extensión of ethological research was the 
application of the comparative method to the phenomena of human cul¬ 
ture. In his 1970 book, Kultur und Verhaltensforschung, Otto Koenig dem- 
onstrated that historically induced, traditional similarities on the one 
hand, and, on the other hand, resemblances caused by parallel adapta- 
tion—in other words, the reciprocal action between homology and anal- 


Introductory History 

ogy—are interacting in the development of human cultures in much the 
same manner as in the evolution of species. For an understanding of cul¬ 
tural history, the analysis of homology and analogy is obviously of the 
greatest importance. 

As a later development of ethology, I should like to mention the con- 
sequences of my own sallies into the field of the theory of knowledge. 
When a stroke of chance shifted me onto the chair of Immanuel Kant in 
Kónigsberg, I was forced to come to terms with Kantian epistemology. 
To anyone familiar with the facts of evolution, the question concerning 
what Kant himself would have thought of the a priori must obtrude 
itself. That is, if he had known about evolution, what would he have 
thought about everything that is given us without previous experience, 
and must, indeed, be given to us in order to make experience possible at 
all? From the viewpoint of the history of Science, it is by no means aston- 
ishing that at least three people not only asked this very question at the 
same time, but also simultaneously and independently of one another 
found the same answer: Sir Karl Popper, Donald Campbell, and I myself. 

In a textbook on basic ethology, I need not necessarily be concerned 
about theory of knowledge, but I think it advisable to attach to the 
English versión of this book an appendix containing a few words about 
the nature of man and about the nature of man's cognitive functions as 
they appeared to me on the basis of my critique of Kantian epistemology. 
This critique seemed controversial during the long-gone days of my pro- 
fessorship at Kónigsberg University. Since ethology has—and especially 
I myself have—so very often been accused of underestimating the dif- 
ferences between man and all other living creatures, I feel justified in 
mentioning, as one of the latest steps in the development of our branch 
of Science, the full recognition of just how different man is from all other 
animáis. Therefore, a short concluding chapter dealing with the unique- 
ness of man will be appended to the last part of the English versión of 
this book. 

Part One 


Chapter I 

Thinking in Biological Terms 

1. The Differences Between the Goals of Physical and 
Biological Research 

In physics the search is for the most general laws governing all matter 
and all energy. In biology the attempt is to understand living systems as 
they are. Since the time of Galileo, physics has proceeded by using the 
method of the generalizing reduction. A physicist always considers the 
individual system he happens to be examining at the moment—it could 
be a planetary system, a pendulum or a falling stone—as a special case 
within a superordinated class of systems. In the examples cited, this is a 
system comprising mass within a gravitational field. Then the physicist 
proceeds to find the lawfulnesses prevailing in one of the special sys¬ 
tems, such as the Keplerian laws and the laws governing pendulums, 
and tries to relate these to the more general laws of the superordinated 
class of systems; in our example, to the Newtonian laws. For this purpose 
he must, naturally, investígate the structures of the special systems. 
Among other things, he must consider the mechanics of the pendulum— 
the axis, the length of the pendulum rod, the weight of the pendulum. 
But, for the physicist, understanding the structures and functions of spe¬ 
cial systems is only a means, only an interim goal on the way toward an 
abstraction of more general laws. As soon as this abstraction is achieved, 
the attributes of the special systems are no longer of any interest at all. 
The individual characteristics of the solar system through which Newton 
discovered the laws of gravitation are completely irrelevant to the valid- 
ity of those laws. He would have arrived at an abstraction of the same 
laws had he contemplated a completely different solar system, a system 


I. Thinking in Biological Terms 

of celestial bodies having different dimensions, intervals, and 

Without exception, the process of the generalizing reduction is 
intended to show how the structure of the special system determines the 
form in which the more general laws take effect in the system in ques- 
tion. The physicist must show, in the case of a pendulum, how the falling 
weight, fixed to a rod of constant length revolving upon an axis, is forced 
into an orbit and, after reaching the lowest point, must move upward 
again within the same orbit; that the frequency of oscillation is deter- 
mined by the length of the pendulum rod and the weight of the pen¬ 
dulum; and so on. 

On the basis of their knowledge of general physical laws, theoretical 
physicists can also, in principie, deduce which special lawfulnesses pre- 
vail in a particular mechanism, for instance in a pendulum. However, in 
the history of physics just as in the history of Science as a whole, it was 
more often than not a case of the investigator stumbling upon a real pen¬ 
dulum before he began to think about its lawfulnesses. Because of this, 
as might be expected, he was also faced with conditions that did not fit 
neatly into his attempts at abstraction. The rod of a real pendulum is 
neither weightless ñor free of inertia; the axial part is not free of friction. 
Yet all these very real facts are for the physicist, at least initially, only 
obstructions which he certainly must take into consideration and, in some 
cases, even measure, but which do not enter into the formulation of the 
law that is finally abstracted. The system a physicist examines during his 
search for general laws holds, in itself, no interest for him, and even less 
interesting is the system's structure. As already stated, understanding a 
system's structure is essentially only an interim goal along his research 
route. He moves, quite literally, by and beyond this as he progresses 
downward and farther downward, to more general and still more gen¬ 
eral laws, down to the conservation principies of physical law. 

Investigators of living systems employ the same method, but with cer- 
tain restrictions to its applicability—a theme I will return to later. But in 
their pursuit of knowledge the goal is not that of the physicist: the biol- 
ogist desires to learn to understand the living system, even if it is only 
a partial system, in itself and for its own sake. All living systems interest 
him equally regardless of their levels of integration, their simplicity or 
their complexity. Just as in the analysis performed by the physicist, the 
biologist also proceeds from the "top" toward the "bottom," from the 
more particular to the more general. We biologists are also convinced 
that one single set of more general and more special laws, in itself devoid 
of contradiction, is ruling the universe. Of these laws the more special- 
ized ones can, in principie, be reduced to the more general ones provided 
one knows the structure of the matter in which they prevail as well as 
the historical génesis of these structures. 

That this second requirement very often fixes limitations for our 

2. The Limits of Reduction 


attempts at reduction is something we will discuss further on. This not- 
withstanding, our endeavors to understand living matter would be 
meaningless if we did not proceed from one basic assumption: Were we 
ever to achieve the utopian goal of completely understanding all life pro- 
cesses, including those transpiring within our own brain—as far as they 
are accessible to objective physiological research—we should be able to 
explain them in terms of the most general laws of physics, provided, of 
course, that we also possess complete insight into the colossally complex 
organic structures that prescribe the various and unique forms in which 
these laws take effect. As investigators of behavior, we hope in the end 
to trace the phenomena which we study back to physical and Chemical 
processes such as those that take place at the synapses, among the elec- 
trically charged cell membranes, and in the conduction of excitation. We, 
too, are "reductionists," although we forget neither that organic life has 
a history ñor that the body-mind problem is insolvable. 

Biologists would be interested in the structures of living things, and 
for their own sakes, even if physiology were not so inextricably inter- 
meshed with pathology. Whoever in day to day work must constantly 
cope with living systems, whether a farmer, a zoo director, a physiologist 
or a physician, cannot help but be involved with disturbances in the func- 
tions of living systems. One repeatedly comes up against the indissoluble 
connection between physiology and pathology for, in order to be able to 
correct a malfunction, one first must understand the normal function 
and, inversely, it is almost always a malfunction which leads to an under¬ 
standing of the normal function. The chances of success are negligible if 
a therapeutic intervention is not guided by insight into the normal func¬ 
tion of the system and into the nature of the disturbance. This is just as 
true when the carburetor of an automobile is clogged as when it is a case 
of human illness. 

2. The Limits of Reduction 

In his significant paper on the role of theory in Science, "Scientific 
Reduction and the Essential Incompleteness of All Science," (1974) Karl 
Popper discusses the successful and unsuccessful attempts at reduction 
undertaken within various branches of Science, and he demonstrates that 
even the most successful of these to date, the reduction of chemistry to 
atomic physical processes, has a remainder, a residue that is impervious 
to further reduction. Popper writes: 

In the course of this discussion, I will defend three theses. First I will suggest 
that scientists have to be reductionists in the sense that nothing is as great 
a success in Science as a successful reduction (such as Newton's reduction— 
or rather explanation—of Kepler's and Galileo's laws to his theory of grav- 


I. Thinking in Biological Terms 

ity). A successful reduction is, perhaps, the most successful form conceivable 
of all scientific explanations, since it achieves what Meyerson (1908, 1930) 
stressed: an identification of the unknown with the known. Let me mention 
however that by contrast with a reduction, an explanation with the help of 
a new theory explains the known—the known problem—by something 
unknown: a new conjecture. 

Secondly, I will suggest that scientists, whatever their philosophical atti- 
tude towards holism, have to welcome reductionism as a method: they have 
to be either naive or else more or less critical reductionists; indeed, some- 
what desperate critical reductionists, I shall argüe, because hardly any major 
reduction in Science has ever been completely successful: there is almost 
always an unresolved residue left by even the most successful attempts at 

Thirdly, I shall contend that there do not seem to be any good arguments 
in favour of philosophical reductionism, while, on the contrary, there are 
good arguments against essentialism, with which philosophical reduction¬ 
ism seems to be closely allied. But I shall also suggest that we should, never- 
theless, on methodological grounds, continué to attempt reductions. The 
reason is that we can learn an immense amount even from unsuccessful or 
incomplete attempts at reduction, and that problems left open in this way 
belong to the most valuable intellectual possessions of Science: I suggest that 
a greater emphasis upon what are often regarded as our scientific failures 
(or, in other words, upon the great open problems of Science) can do us a lot 
of good. 

Then Popper cites a series of examples which illustrate how becoming 
bogged down during attempts at reduction has led to the discovery of 
previously unrecognized problems. According to Popper, this was 
already evident in mathematics and the attempt at "arithmetization;" 
that is, the reduction of geometry and the irrational numbers to rational 
sums did not lead to complete success. Popper States: 

But the number of unexpected problems and the amount of unexpected 
knowledge brought about by this failure are overwhelming. This, I shall 
contend, may be generalized: even where we do not succeed as reduction¬ 
ists, the number of interesting and unexpected results we may acquire on 
the way to our failure can be of the greatest valué. 

Another example of the same principie and especially important for 
our purposes is aíforded by the attempt to reduce chemistry to quantum 
physics. Even if one chose to assume that it could be possible to reduce 
the nature of the Chemical bond to principies of quantum physics— 
something which has not yet been achieved—and even if, for the sake 
of argument, one chose to assume that one had in hand thoroughly sat- 
isfactory theories of nuclear forces, of the periodic system of elements 
and their isotopes, etc., the attempt to reduce Chemical processes to quan¬ 
tum mechanics would still be stopped by a barrier, by an idea funda- 

3. Ontological Reductionism 


mentally foreign to the way a physicist thinks and to the body of phys- 
ical theory: the historical concept of becoming, of génesis. Bohr's new 
theory of the periodic system of elements assumes that the heavier nuclei 
have been formed out of lighter ones, in other words, they carne into 
being through a prior historical process in which, in thinly scattered and 
exceedingly uncommon cases, hydrogen nuclei fused into heavier 
nuclei—something that can occur only under conditions which, appar- 
ently, are very rarely encountered in the cosmos. Physicists have strong 
indices in favor of the view that this really happened and still happens. 
Beginning with helium, all the heavier elements are the result of cos- 
mological evolution. "Thus," says Popper, "from the point of view of 
method, our attempted reductions have led to tremendous successes, 
even though it can be said that the attempted reductions have, as such, 
usually failed." 

3. Ontological Reductionism 

The assumption on which all Science is based, that everything existing 
carne into being through combinations of material elements in the course 
of a great cosmic creation and, even to this day, is still governed by those 
same laws which prevail in these elements, can easily lead to a philo- 
sophical error that is dangerous because it tends to discredit all natural 
Science in the eyes of sensible, thinking people. This error arises, as do 
so many others, through the neglect of the structures within which the 
omnipresent laws of physics opérate in the form of highly complex spe- 
cial laws. These, however, have every right to be designated as natural 
laws and as having the same dignity as the first law of physics or the 
laws of conservation. 

The consequence of this error is the expression, everything that is com- 
posed of matter is "nothing else but" this matter. Although this "nothing 
else but" has been shown to be erroneous at even the most basal physical 
levels, great scientific innovators have continued to cling to it with 
remarkable tenacity. Popper (1974) presents the example of Eddington 
who, for a long time, continued to believe that with the advent of quan¬ 
tum mechanics the electromagnetic theory of matter had been finally set- 
tled and that all matter consisted of electrons and protons—at a time 
when the mesón had already been discovered. 

Ontological reductionism becomes really dangerous when applied to 
the subject of living organisms. If, for example, we say: "All life processes 
are Chemical and physical processes," anyone accustomed to thinking in 
scientific terms would not question the correctness of the statement. But 
if we say: "All life processes are essentially only Chemical and physical 
processes," every biologist would protest, since it is with regard to just 


I. Thinking in Biological Terms 

this, to what they essentially are, that life processes are different from all 
others: they are not just chemical-physical processes. The failure of 
ontological reduction becomes even more conspicuous when we com¬ 
pare two other statements: "Humans are beings of the class of mammals 
and the order of primates," and "Human beings are essentially nothing 
else but mammals of the order of primates." 

The "nothing else but" of ontological reductionism, for which Julián 
Huxley coined the term "nothingelsebuttery,"* aptly describes a blind- 
ness to two most essential realities: first, to the complexity of organic 
structures and the various levels of their integration, and second, to those 
sensibilities of valué which every normal human being extends toward 
the lower and higher achievements of organic génesis. 

During evolution, new systemic properties often arise through the 
integration of subsystems which, up to that moment, had been function- 
ing independently of one another. A simple electrical model (Figure 1) 
taken from a book written by Bernhard Hassenstein (1966) can illustrate 
how, through such an integration, completely new systemic properties 
can be generated instantaneously, properties which did not exist before 
and, most important, also gave no prior indication of existence. Herein lies 
the truth of the mystical-sounding but nonetheless completely accurate 
statement made by Gestalt psychologists: "The whole is more than the 
sum of its parts." Through cybernetics and system theory the sudden 
emergence of new systemic characteristics and functions has been 
accounted for by means of a purely physical approach and those who 
have done the research and written the descriptions have thus been 
freed from any suspicion of a vitalistic belief in the miraculous. 

Philosophers not familiar with this characteristic process of evolution 
tend to believe that all evolutionary changes are accomplished in slow, 
gradual transitions. As a result of this belief there are frequent ontolog¬ 
ical disputes about whether, for the organism in question, the difference 
between two evolutionary stages is one of degree or one of essence. 
Understandably, this dispute rages most vehemently about that evolu¬ 
tionary stage where the step from animal to human being was taken. 
Mortimer J. Adler has dedicated an entire book, The Difference of Man and 
the Difference It Makes (1967), to a discussion of this question. In reality, 
almost every larger evolutionary step signifies the emergence of a dif- 
ferentiation of essence since it consists in the appearance of something 
that had never existed before. This is also true, in principie, even for the 
differentiation in Hassenstein's inorganic systems which were brought 
in as illustration. 

* I was introduced to this term by my friend, Donald Mackay, but my much older 
friend, Sir Julián Huxley, was able to demónstrate a priority for this beautiful expression. 

3. Ontological Reductionism 


Figure 1. Three electric circuits, among them an oscillating Circuit (c), illustrate 
the concept of "systemic property." A current is passed between the poles of a 
battery with electromotive strength e 0 and with the potential tensión or terminal 
voltage V 0 . The resistance of the Circuit, in ohms, is concentrated in R. Circuit (a) 
has a condenser with capacity C; circuit ( b ) has a coil with inductance L; Circuit 
(c) has both condenser and coil together. The voltage V can be measured at the 
two termináis. The diagrams at the right represent the changes in voltage after 
the Circuit is closed. In circuit (a) the condenser gradually charges through the 
resistance until voltage V G is reached. In circuit (b) the current, initially impeded 
by the self-induction of the coil, increases until it reaches the strength laid down 
by Ohm's law; the voltage V is then theoretically zero because the total resistance 
is concentrated in R. In circuit (c), closing the circuit causes diminishing oscilla- 
tions. It is apparent that the systemic properties of (c) are not the result of super- 
imposing the properties characteristic of (a) and (b) although (c) can be thought 
of as having arisen through a combination of (a) and (b). The diagram is valid, for 
example, for the following valúes: C = 0.7 X 10 -9 F; L = 2 X 10~ 3 Hy; R = 10 3 
o?; X = 1.2 X 10 -6 sec. This last valué also defines the time axis, which is the same 
for all three curves. Calculations by E. U. von Weizsácker. (Hassenstein, B.: Kyber- 
netik und biologische Forschung. Handb. der Biologie I. Frankfurt: Athenaion, 1966.) 


I. Thinking in Biological Terms 

4. The Evolutionary Event as a Limitation of Reduction 

As we have seen, the insurmountable obstacle of the historical event has 
impeded the endeavors to reduce all of chemistry to quantum physics. In 
order to push reduction beyond the existence of the heavy nuclei, that 
is, in order to explain why these are thus and not otherwise, one would 
need insight into the process of cosmogeny which produced them. Only 
rarely does the physicist run up against this limitation; the biologist does 
it each step of the way. 

Literally every characteristic detail of form and function of every liv- 
ing organism is determined by the succession of evolutionary events 
which engendered that organism. This historical progression is, in turn, 
determined by an astronomical number of causal chains which converge 
into its channel so that, in principie, it is impossible to trace any of them 
back for more than a very short way. As I show in the next chapter, the 
route evolution has taken can be reconstructed to a modest extent, but 
not the causes which determined that route. In his attempts at reduction 
and explanation, the biologist must reconcile himself to a very large res- 
idue that cannot be rationalized historically. He can explain the func- 
tions he observes in his object of study only through its structures. This 
forces him to focus his attention on those structures, not for themselves 
alone but also for the sake of their functions. 

The biologist who attempts to reduce what is there to lawfulnesses 
already known, or to explain what exists through new hypotheses, is 
confronted with two tasks which, although they overlap and merge, can 
still be clearly differentiated. The first task is an analysis of the form and 
function of the living system as it is at this moment in evolutionary time; 
the second task is to make understandable how this system got to be that 
way and no other. The duality of this problem can best be illustrated by 
using a man-made machine as an example. If a combat airplane built by 
the Russians were to fall into the hands of the Japanese, they would take 
it apart, they would analyze it in a literal sense, and most likely this anal¬ 
ysis would succeed to such an extent (still utopian for living systems) 
that they would be capable of complete resynthesis, that is, they would 
be capable of constructing a replica of the airplane. But they would also 
certainly find quite a few details of construction that were ingeniously 
thought out and which would cause them to wonder how in all the 
world the designer had hit upon this original solution. The prerequisite 
for finding an answer to the question why the foreign designer had 
devised a particular part in such a way and not in another would be, to 
begin with, not only an immense amount of information about the his- 
tory of Russian aircraft construction, but, over and above this, also about 
the thought processes, the brain and nerve physiology of the designer, 
and so on and so forth back into the entire phylogeny of Homo sapiens. 

5. The Question "What For?" 


That the two tasks can also be undertaken separately for the exami- 
nation of a living System is confirmed by the fact that analysis of living 
Systems as they are at the moment was considerably far advanced before 
questions concerned with how they had gotten that way were ever 
asked. Harvey had clearly comprehended the function of the circulatory 
System and had consequentially postulated the existence of capillaries, 
which one could not even see at the time. An entire series of discoveries, 
made independently of any historical consideration of organic génesis, 
is a part of the past of the Science of medicine. 

For a long time, long before the main causes of all adaptation (genetic 
change and natural selection) had been discovered, those who worked 
with living Systems had been aware that their constructions were con- 
stituted, in general and in more special and improbable ways, to sustain 
the life of the individual, its progeny, and, consequently, its species. 
Those who believed in a singular act of creation explained this species- 
preserving purposefulness either through the wisdom of the creator or, 
as Jakob von Uexküll (1909), who rejected evolutionary ideas, through a 
prestabilized harmony between the organism and its environment. 
Those who can be contení with these explanations are spared the asking 
of many difficult questions but, it must be remembered, at the price of 
having to cióse their eyes to certain phenomena—and these are the 
many characteristics extant in the construction of living organisms that 
are inexpedient for survival. 

5. The Question "What For?" 

At this point, the historical barrier of reduction forces the researcher to 
ask a question with which, up to this point in his attempts at reduction 
and explanation, he has not been confronted: he is forced to ask about 
purpose. Processes which are determined by an end or a goal exist in the 
cosmos exclusively in the realm of organisms. According to Nicolai Hart- 
mann (1966), a goal-directed action can only be understood on the basis 
of insight into the interaction between three processes: First, a goal must 
be set and in this setting something that will happen in the future must 
be anticipated by "skipping" spaces in time. Second, the choice of means, 
dictated by the set goal, must be made. Third, the realization of the set 
goal, through the causal sequential unfolding of the chosen means, is 
achieved. These three acts form a functional unit and are performed at 
various levels of integration within the realm of the organic (Figure 2). 

Nicolai Hartmann believed that the actor and the goal setter could 
never be anything but a consciousness for, as he said, " . . . only a con- 
sciousness has maneuverability within conceptual time, can leap beyond 
sequential time, can predetermine, anticipate, choose means, and retro- 


I. Thinking in Biological Terms 

Figure 2. Hartmann's conceptualization of the processes underlying goal-directed 
action. (Hartmann, N.: Teleologisches Denken. Berlín: Walter de Gruyter, 1964.) 

spectively go back in thought over the skipped spaces through to the 
first one at the beginning." Since Hartmann wrote this statement, 
research into the biochemistry of morphogenesis and also into that of 
appetitive behavior in animáis (Two/I/10) has revealed processes in 
which the three acts required by Hartmann are performed in distinct 
interaction and yet in sequences that are certainly not dependent on or 
associated with consciousness. 

If the preexistent "blueprint" in the genome anticipates the construc- 
tion of a new organism as a goal, and if, subsequently, this goal is 
attained through a quite variable and adaptive choice of those means 
proffered by the milieu and within a strictly causal sequence of devel- 
opmental steps, this then conforms to the combined functions of the 
three acts postulated by Hartmann. The same is true for the admittedly 
somewhat more complicated processes of appetitive behavior through 
which an animal achieves a goal meaningful for species preservation, 
which are also programmed in the genome. Although these processes of 
embryogenesis and those of behavior are certainly not governed by an 
anticipatory individual consciousness, but are enacted at a very much 
lower level of organic occurrence, they must still be regarded as com- 
pletely purpose-oriented or "determined by their ends or goals." 

The processes of phylogeny, which generate the "blueprints" under 
dicussion, accomplish truly marvelous things, yet they lack completely 
the infallibility of an all-knowing, purpose-setting creator as well as the 
constancy of a prestabilized harmony. All along the way, the compara- 
tive phylogeneticist meets "errors" of evolution, faulty constructions of 
such shortsightedness that one would not credit them to any human 
engineer. Gustav Kramer cited many examples of this phenomenon in 
his publication on the inexpedient in nature (1949). Just one of these 
examples will suffice here. During the transition from aquatic life to ter- 
restrial life the swim bladder of the fish became an organ for breathing. 
In the circulatory system of fish, and already evident even among the 
jawless cyclostomes (Cyclostomata), the heart and gills are coupled in 

5. The Question "What For?" 


tándem, that is, all the blood pumped from the heart must, of necessity, 
pass through the gills, and from there the richly oxygenated blood is 
circulated throughout the body. Because the swim bladder is an organ 
serviced within the body's system, and has remained so even after 
becoming an organ of respiration, the blood emerging from it flows 
directly back into the heart. Henee the heart is the recipient of oxygen- 
poor blood from the body and oxygen-rich blood from the lungs and a 
mixture of these is what is continually being recirculated. Although from 
a technical point of view this is a most unsatisfactory solution, it has been 
maintained through many geologic eras and up to this day by all 
amphibians and almost all reptiles. It may be because of this kind of con- 
struction of the vascular system that none of the animáis just mentioned 
is as capable of sustained muscular exertions as are many sharks (Sela- 
chii) and bony fishes (Teleostei), as well as all birds and mammals. 

The fact that individual development, the ontogeny of every living 
creature, represents a genuine goal-directed process, the realization of a 
preexistent plan, must not lead to the altogether erroneous conclusión 
that the same is true for evolution. This error, unfortunately, is suggested 
by the very words "development" and "evolution." 

Evolution is emphatically not purpose-oriented in that it is absolutely 
unable to foresee what might be expedient in the future. Evolution is 
incapable of accepting even the smallest disadvantage for the sake of a 
future advantage. In other words, evolution can exploit only those fac- 
tors which tender an immediate "selective advantage," somewhat in the 
same manner that even the most farsighted and benevolent politician is 
confined to the exploitation of those issues and measures which are 
capable of providing him with an immediate "electoral advantage." 

Thanks to some very oíd discoveries made by Charles Darwin, we 
know which processes of mutation and natural selection provided organ- 
isms with sufficient purposefulness (in the sense of species preservation) 
and, thanks to some new knowledge gained through molecular genetics, 
we now know fairly precisely how the information underlying all adap¬ 
taron is coded in the "blueprint" of the genome. Manfred Eigen has con- 
vincingly demonsrated that selection, the natural choice of the most suit- 
able, is operative even at the molecular level and that it played a decisive 
role in the origin of life. But the stuff on which selection gets to work is 
always only the purely happenstance alteration or new combination of 
heritable characteristics. It is technically correct to say that evolution pro- 
ceeds only according to the principies of blind chance and through the 
elimination of what is incapable of sustained existence, and yet this way 
of expressing it is misleading. The fact, indisputable in itself, that the 
few billion years of existence reckoned for our planet by radiation 
researchers should have, in this way and by these means, sufficed for the 
evolvement of man from pre-life forms similar to the most simple blue- 
green algae (Cyanophyta), seems incredible to those not conversant with 


I. Thinking in Biological Terms 

all that is involved. The wildness of chance is, in these circumstances, 
"tamed"—as Manfred Eigen (1975) expresses it—by the benefit it brings 
on very rare occasions. The likelihood of a mutation which improves the 
organism's chances of survival is extremely small; in fact, the probability 
is assessed by great geneticists as amounting to 10~ 8 . The improbability 
of such a happy event is compensated by the correspondingly great 
increase of the probability of survival that is enjoyed by the "lucky 

If the genetic equipment of an individual increases his chances of sur¬ 
vival relative to those of his conspecifics, this is then so only for that 
unique constellation of environmental factors with which this creature 
must actually contend. For a different kind of environment, the newly 
acquired genetic equipment could very well be inexpedient. If, in the 
course of time, the environment of an animal species changes (climatic 
changes, for example, or new conditions concerning food sources, or an 
increase in the number of predators), then the average genetic equip¬ 
ment transmitted by individuáis of the species is usually no longer opti- 
mally adapted. Selection, consequently, supports some of the genetic var- 
iations available in stock and suppresses others, so that after a 
corresponding period of time, a new set of genetic equipment better 
adapted to the incursive conditions is achieved. One can describe this 
process with the help of communication concepts and by means of the 
following formulation: Through the process of selection, new data con¬ 
cerning the conditions of the external environment reach the genome 
and are henceforth stored there as "instructions" (Eigen 1975). With 
regard to this acquisition of environmental information, the process of 
evolution resembles the cognitive process of the acquisition of knowl- 
edge. This knowledge, however, like invested capital, pays such high 
"dividends" that it compensates for the rareness of advantageous genetic 

To quote Eigen once again: "Life is a game in which nothing is stipu- 
lated but the rules." The family tree of all living things is a typical exam¬ 
ple of what game theorists cali a "tree of decisions." Eigen chose an aerial 
photograph of the Colorado River delta (Figure 3) as an actual example 
representing such a tree: billions of unknowable causal influences deter¬ 
mine which course every single rivulet and channel has taken. 

If one knows a little bit about the rules of the game being followed in 
the growth of the family tree of life, one is no longer astonished to find 
in the structures of living organisms not only the constructive Solutions 
suggesting genius but also those that could be appreciably improved by 
any human engineer. No plumber, for example, would lay the pipes and 
tubes for the blood vessel system of a lizard so impractically as evolution 
has done it. The principies of mutation and selection make it understand- 
able that in the realm of the organic we meet not only characteristics of 


I. Thinking in Biological Terms 

structure and function which are practical but also those which are not 
so impractical that they lead to the elimination of the bearers. 

One other important characteristic of all living organisms, readily 
understandable through some knowledge of the ground rules of the evo- 
lutionary game, is the extreme "conservatism" of their structures. But 
through a change in an organism's way of life, especially when adapta- 
tions are required to a new environmental niche, oíd structural features 
can become meaningless. As long as their presence does not directly 
threaten the continued existence of the species in question, they can be 
lugged along almost indefinitely, and this happens if their retention does 
not result in a loss of too many individuáis of the species. Before the 
development of modern surgery, for example, thousands of humans per- 
ished every year from peritonitis because an organ which had become 
funtionless, the vermiform appendix, a narrow blind tube extending 
from the caecum, tended to become inflamed. 

In many cases, structures which have become functionless are con¬ 
verted to serve other useful functions, simply because they are there. 
From a gilí cleft, the spiracle (Spiraculum) of sharks and primitive fishes, 
comes the ear of higher vertebrates; from the first gilí arch, the hyoid 
bone with its two pairs of branches; and so on. 

Thus the over-all construction of an organism is never comparable to 
a human edifice that has been designed in a unique process of planning 
by farsighted architects and engineers. An organism's structure is much 
more like the do-it-yourself house of a homesteader who first puts 
together a simple log cabin as protection against the wind and rain and 
then, as the extent of his holdings and the number of people in his fam- 
ily increases, periodically enlarges his dwelling. The original cabin is not 
torn down; instead, it is used as a storeroom, and almost every room of 
the expanding house will, in the course of the structure's development, 
be used for some purpose other than that for which it was originally 
intended. The older parts of the house, recognizable for what they are, 
are retained because the evolving structure could never be entirely dis- 
mantled and newly planned; this is not possible because every part has 
been continuously occupied and intensively used. 

In an analogous way, characteristic structural details that are retentions 
of "yesterday's adaptations" can be found in every organism. Their exis¬ 
tence is a stroke of really good luck for the researcher who wants not 
only to learn the game rules of evolution but who also wants to know 
the path along which evolution actually has proceeded. 

All of the preceding paragraphs have been written with the intention 
of explaining to those who are not biologists what is meant when a 
biologist asks the question, foreign to physicists, "What for?" If we ask 
"What does a cat have sharply pointed, curved and retractile claws for?" 
and answer straightaway "To catch mice," this question and answer do 

5. The Question "What For?" 


not signify in any way an acknowledgment of an inherent purposive 
orientation of the universe and of organic evolution. It is, rather, an 
abridgment of the question, "Which is the particular function whose sur- 
vival valué has bred cats (Felidae) with this form of claw?" Colin Pitten- 
drigh (1958) has designated this way of regarding species-preserving 
purposefulness as teleonomy in order to draw as sharp a distinction 
between it and a mystical teleology as exists today between astronomy 
and astrology. 

In the structural plan of an animal—more than in that of a plant— 
hardly any characteristic exists that is not in some way influenced by 
selection and is, in this sense, thus, teleonomic. If I were to search for 
examples of purely happenstance characteristics and characteristic diver- 
sifications, it would be difficult to cite any occurring among wild animáis. 
Certainly it is teleonomically irrelevant if a barnyard chicken of no rec- 
ognizable breed is white and another brown or mottled. Among wild 
animáis, the African hunting dog (Lycaon pictus L.) is one of the few 
examples of a species whose individuáis exhibit variable dappled pat- 
terns though in fairly uniform colors. But even here, in the highly vari¬ 
able color patterns of these wild animáis, the question arises if this var- 
iability, as such, could not also be teleonomic and thereby the result of 
selection. It is certainly possible that, among these extraordinarily social 
predators, if individual animáis are able to recognize one another even 
at great distances, this capability can have valué for the preservation of 
the species. The balanced dimorphism of male ruffs (Philomachus pugnax 
L.) can, with assurance, be explained in this way. 

When dealing with such superficial physical characteristics it is partic- 
ularly difficult to find any which demonstrably have nothing to do with 
teleonomy. Within complex configurations of various characteristics 
there are probably none for which patient research could not currently 
accumulate factual confirmation that its explicit function was selected in 
just this way because of its species-preserving valué. The more complex 
and generally more improbable such a combination of characteristics is, 
with that much more certainty can one conclude from these a relation- 
ship between function and selection, and that much more easily answer 
the question, "What for?" 

In a paradoxical way, characteristics which apparently disturb a partic¬ 
ular function generally important for survival are exactly those which 
provide us with the most certain answers to the question, "What for?" 
Citing only one example, the fish Heniochus varius, classified among the 
Chaetodontids, has two horns just above its eyes and above these, on its 
nape, another thick, rounded horn; between them there is a deep saddle- 
like indentation (Figure 4). These structural characteristics are obviously 
cumbersome, certainly not streamlined and, for slipping through narrow 
passages—an activity that belongs to the daily routine of this coral-reef 


I. Thinking in Biological Terms 

Figure 4. Heniochus varius Cuvier. The horns above the eyes and the hump on the 
upper back serve in ritual fighting; this fighting certainly developed from the 
damaging fighting of the related Chaetodontids who spear their rivals with the 
spines of the anterior dorsal fins. 

fish, most disturbingly in the way. The argument that these disadvan- 
tages must somehow "pay off" through some kind of selective valué is 
compelling. In fact, when I first got one of these fish, I predicted the 
function of this saddle that is bordered above and below by horns: The 
species would engage in ritual fighting during which opponents, when 
confronting one another, would lie aslant, both leaning either to the 
right or to the left so that the saddles fitted into one another, and then 
the combatants would shove against one another with all their strength. 
Our first "What for?" is, thus, satisfactorily answered with the reply, "For 
ritual fighting." What ritual fighting should have a selection premium 
for in a species of fish that normally swims around in large schools is 
another question, and the answer to this question I hope to reach 
through further observation. The prediction is that it is bred by intraspe- 
cific selection. 

Adaptations like the ones just cited, those which serve exclusively for 
combat with conspecifics, are by no means always conducive to species 
preservation. The teleonomic query, "What for?" does, in fact, get a clear 
answer since the valué for selection that the behavior or structure in 
question has developed for the genome of the individual is easy to dem¬ 
ónstrate. But whether the connection in question is advantageous for the 
preservation of the species as a whole is not at all certain. Quite the con- 
trary, "intraspecific competition" among rival conspecifics can easily lead 
to an escalation of extremes in forms and functions, extremes which are 
not conducive to the continuation of that species. Those specialized wing 
primaries of the Argus pheasant (Argusianus argus) enhancing the effect 
of some particular courting movements are no longer much good for 
flying. Yet the larger they are, the greater are the chances for the cock 
possessing them to foster numerous offspring—but, at the same time, the 
greater are his chances of being eaten by ground predators. As always. 

5. The Question "What For?" 


life's forms and their functions are a compromise between several kinds 
of selection pressure. 

Understandably, the selection pressure exerted by one individual of a 
species on other individuáis of the same species does not provide any 
information about which interactions with its environment are advan- 
tageous or disadvantageous for that species. When deer (Cervidae) "take 
a fancy to" a particular kind of ritual fighting whose outcome depends 
on the size of the antlers growing out of the foreheads of the males, and 
when, in addition to this—as Bubenik (1968) has shown for the red deer 
(Cervus elaphus )—females actively choose partners according to the size 
of these antlers, there is nothing to hinder the bulk of these osseous trees 
(which must, moreover, be grown anew each year) from increasing to 
the limits of what can be carried. 

Among some mammals living in social groups, intraspecific selection 
impels the appearance of phenomena that one could regard almost as 
deformities; but these incorpórate the same principie as that of the deer 
antier. Among some Asiatic monkeys (Semnopithecinae), each male com- 
mands a large harem of females. Each change of such a commander is 
accompanied by general infanticide within the harem. Because females 
deprived of their offspring come into estrus earlier than those allowed to 
rear their young, the new tyrant has a much greater opportunity for 
transmitting his genome. 'The egoistical gene" has become the popular 
phrase to use when referring to this remarkable phenomenon. It is not 
at all easy to decide to what extent such "egoism" can be carried before 
it harms the entire species and, finally, the gene itself as well. 

With this the difficult problem of kin selection is broached. It does not 
matter which example of pronounced "altruistic" behavior one chooses 
to use, be it food regurgitation among African hunting dogs (Lycaon pic- 
tus L.) or mutual defense of comrades among jackdaws (Coloeus monedula); 
when one asks oneself why a mutation excluding this altruistic behav¬ 
ior—an excluding mutation that must certainly occur from time to time 
and, when it does, must obviously be selectively advantageous for that 
individual—is not at once and prolifically further selected for, one finds 
no answer. Questions concerning those factors which prevent a massive 
emergence of social parasites among animal societies remain to be 
answered adequately. It is nothing short of astonishing that, among com- 
munal vertebrates, social parasitism is as good as unknown and infanti¬ 
cide, to date, known to occur only among a few species. Among higher 
and longer-lived Metazoa, at the level of entireties of cellular organisms, 
there is a highly complicated system of antibody formation that exists so 
that asocial elements can be eliminated whenever this becomes neces- 
sary. At the level of human culture, no single ethnic group is known 
which does not have comparable, complex Systems of laws and tabus for 
the suppression of every kind of antisocial behavior. Among animal 
societies, which can be classified at a level of integration somewhere 


I. Thinking in Biological Terms 

between human society and the bodily entirety of Metazoa, no mecha- 
nisms for the prevention of social parasitism are, as yet, known. 

As can be seen, the teleonomic question, "What for?" often elicits 
responses that indícate goals that are rather narrow-minded and short- 
sighted; for species preservation, the results of intraspecific selection 
often are decidedly detrimental. 

Answers are apparently not forthcoming to the teleonomic question 
where structures and even behavior patterns have changed their former 
functions. "Yesterday's adaptations" are to be found everywhere in mor- 
phology and in behavioral Science. Structures which have become mean- 
ingless are seldom simply disassembled; instead, they are almost always 
put to use serving a function other than that for which they were ini- 
tially intended: a gilí aperture becomes an auditory canal; an organ 
which served for the collection of nutriment becomes an endocrine 

6. Teleological and Causal Views of Nature 

Among the two great schools of the behavioral Sciences, only those sci- 
entists identified with the school of purposive psychology—out-and-out 
teleologically and in no way teleonomically oriented, and represented 
by distinguished researchers such as William MacDougall (1923), Eduard 
C. Tolman (1932) and others—concentrated their entire research interest 
on the question, "What for?" In other words, they proceeded as if every 
question would be answered and all problems would be solved by deter- 
mining the goal of any animal behavior. If one has understood that 
flying southward in autumn serves birds of passage as a species-preserv- 
ing end or goal, then, in the opinión of the teleologically oriented animal 
psychologist, Bierens de Haan (1940), all the problems of bird migration 
are solved in a "satisfactory and elegant" way. An enormous amount of 
one-sided thinking is required in order to be able to regard only the 
question of ends and goals as fundamental and to neglect completely 
questions of cause. 

There are cases in which the function of a particular morphological 
structure becomes apparent even after superficial observation. One 
thinks, in this context, of the patterns of vivid color that play a role for 
so many releasing mechanisms. The white collar of the drake mallard has 
been shown to act as a signal during courtship. What constitutes this col¬ 
lar is a differential color distribution among the neck feathers involved. 
These look as if the white had been applied with a paint brush over the 
already completely formed and finished feathers. At the top of the collar, 
near the head, the white begins as a border along the lower edges of the 
green feathers. Farther down and back, the white border gradually 
becomes broader until the feathers are completely white from the base 

6. Teleological and Causal Views of Nature 


to the tip. At the bottom or back of the white collar, the brown color of 
the lower neck feathers begins to appear; the otherwise white feathers 
have borders of brown and these borders broaden until the feathers are 
completely brown. It is very easy to recognize that the white collar which 
separates the green of the head from the brown of the lower neck is 
"intended" to appear sharply delineated. In contrast to this easy recog- 
nition, a reduction of this "artistic" structure to its physiological and 
genetic causes is very difficult to achieve. One must realize how complex 
the morphogenetic processes producing such a structure must be, and 
then, even more difficult, one must try to imagine what the phylogenetic 
processes may be that lead to the blueprint of this morphogenesis. "Cases 
such as these," Otto Koehler (1933) has said, "which at first afford an 
understanding only of goal-directedness, can easily seduce [emphasis 
mine] the research worker into making hasty Lamarckian interpreta- 
tions; these cases are actually there to be regarded as challenges for causal 

Understandably, these "challenges" do not exist for the biologists and 
behavioral scientists who think in vitalistic and teleological terms. For 
them the acceptance of extranatural factors makes it possible, with no 
ado at all, to answer every "What for?" question with a spurious 
"Because." The most unfortunate consequence of this methodological 
and conceptual attitude was its effect on the opposing school: the pur- 
posivists' misuse of goals and ends for what pretended to be an expla¬ 
naron aroused in the behaviorists an emotional opposition to all consid- 
erations of purpose, so that they developed a "blind spot" for the concept 
of species-preserving purposefulness as well. It is now part of history 
that those of the behaviorist school of psychology, motivated by their 
antagonism to teleological purposive psychology, dismissed all consid- 
erations of species-preserving teleonomy—a classic example of how, 
because of antagonism among members of two schools of thought, a 
rejection of knowledge can be occasioned. 

When engaged in research on a complex living system, it is obvious 
that questions concerning ends and goals as well as questions concerning 
causality must be asked simultaneously. I would like to illustrate this by 
means of a parable about a man and a man-made system. The man is 
traveling overland in his automobile; the purpose of his journey is a lee- 
ture which he is scheduled to deliver in a distant city. The man is under- 
way "for lecturing"; his automobile, a means serving the same end, is 
there "for traveling." The man takes delight in contemplating this won- 
derfully graduated and integrated complex entirety of interacting ends 
and goals. Perhaps, at just this moment, he is admiring the phrase, "vítale 
Phantasie," which F. J. J. Buytendijk (1940) used to describe and explain 
the purposefulness of body structure and behavior, something which the 
automobile manufacturer has apparently confirmed in the design and 
planning of this purposeful vehicle. Then something happens that hap- 


I. Thinking in Biological Terms 

pens often; the automobile motor coughs, splutters, and stops. All at once 
the driver is most impressively presented with the fact that the goal of 
his journey is not what makes the automobile move. As I choose to 
express it, goal-directedness does not "exert pulí" in the direction 
desired. Now it would behoove the man in the automobile to forget, for 
the time being, the aim of his journey and to devote his attention to the 
causality of the normal function of the motor as well as its momentary 
malfunction. On the success of his causal investigations will depend the 
possibility of further pursuing the goal of his journey. 

The "goal-directedness" of every systemic whole is not one mite less 
dependent on the causality of its organs than is the goal-directedness of 
the traveling lecturer in our parable on the functioning of his automobile 
motor. With regard to organic systems, the regulating function of causal 
research has an incomparably more difficult task to perform, not only 
because these systems are much more intricately structured than the most 
complicated machine, but also because, in contrast to machines, they 
have not been put together by men and we have only very incomplete 
insight into their origin and the history of their formation. Thus the phy- 
sician who attempts to repair a malfunctioning organic system has that 
much more difficult a task than the automobile mechanic. But the success 
of what he does to regúlate the functioning of a whole system will also 
depend, basically, on his investigation of causality and will be influ- 
enced just as little by the amount of "urgency" that exists for reaching 
the now problematic goal. 

Questions about how the malfunctioning systemic whole originally 
carne into being are fundamentally of no real importance in the relation- 
ship referred to here between its goal-directedness and the mode and 
manner of its causal comprehensibility. If an engineer constructed the 
purposeful system, if a God created it, or if it carne into being by means 
of a natural evolution in which, besides mutation and selection, many 
other processes could also have played a role—all these "ifs" are quite 
irrelevant to the capacity of human comprehension of causality, to an 
insight that can master such a system and, when the proper functioning 
of the system is endangered, can actually restore it. 

The ultimate significance of human research into causality is to be 
found in the fact that this research gives us, as the most important regulat¬ 
ing factor, the means to control natural processes. Whether these processes 
are external and inorganic, such as lightning and storms, or internal and 
organic, such as diseases of the body or "purely psychic" symptoms of 
decline in the social patterns of human behavior is, moreover, of no con- 
sequence. Never is the pursuit of an actively predetermined goal possible 
without causal comprehension. On the other hand, causal analysis would 
have no function if questing humanity did not pursue goals. 

The endeavor "to search, as far as it is possible for us, among the 
world's causalities and to trace the chain they form as far as they are 

6. Teleological and Causal Views of Nature 


linked to one another" (Kant) is not "materialistic" in some moralistic 
sense, as some teleologists choose to describe it, but signifies a potential 
for the most intensive active Service to the ultimate aims of all organic 
processes in that we, where success is achieved, are endowed with the 
power to intervene, helpfully regulating, where valúes are endangered, 
while the purely teleological observers can only lay their hands in their 
laps, deedless, mourning the disintegration of the whole. 

Chapter II 

The Methodology of Biology and 
Particularly of Ethology 

1. The Concept of a System or an Entirety 

The goal of biologists is, as I have said, to make an organic system under- 
standable as a whole. This does not mean that the biologist regards the 
entirety of a system as some kind of miracle. It is necessary to make this 
clear at the very beginning since there are some atomistic theoreticians 
who regard it as a confession of vitalism if one merely utters the words 
"whole" or "entirety." The biologist does not believe in "whole-produc- 
ing factors" that are neither in need of ñor accessible to an explanation, 
but he remains aware that the systemic character of the organism 
exeludes the utilization of some of the less sophisticated research meth- 
ods. Above all, with regard to an organic system, one cannot track simple 
and unidirectional linkages between causes and effeets. In his 1933 mon- 
ograph, "Die Ganzheitsbetrachtung in der modernen Biologie," Otto 
Koehler developed in detail the methods which are necessary in order to 
analyze a systemic entirety. In this publication he grants the Gestalt psy- 
chologists the credit they deserve for having perceived the nature of 
organic entireties, although he justifiably criticizes them where neces¬ 
sary in a way that can be summarized as follows: Every gestalt is an 
entirety, but not every entirety is a gestalt; in other words, the concept 
of gestalt must be reserved primarily for the processes of perception. 
Koehler also appropriately emphasized the fact that impassioned cham- 
pions of the principies of entirety, such as B. H. Driesch (1928), alienated 
many researchers from the theory of entireties because they "dressed it 
in the vestments of vitalism." This I can readily verify; as a young natural 
scientist I, too, was convinced that only atomistic component analyses 
could claim to fulfill the requirements characteristic of an exact Science. 

1. The Concept of a System or an Entirety 


Otto Koehler (1933), together with Ludwig von Bertalanffy (1951), ren- 
dered a great Service by showing that an organic system requires a special 
analytic methodology of its own, even if one were to undertake an anal- 
ysis that was intended to be purely physical. There is, said Koehler 
(1933), no "vital forcé," and no "whole-producing factor"; there are, 
"instead, harmonious Systems of amboceptor [that is, operative in both 
directions] causal chains, entangled and integrated, whose harmonious 
interactions are just what constitutes the configurational entirety." 

This does not imply, by any means, that an entirety is fundamentally 
inacessible through causal analysis. Koehler (1933) said: 

Everyone who does research—and all of our research is primarily analysis— 
is consciously 'simplistic', and must be so in order to maintain his own 
momentum. To this extent it is understandable if just those particularly suc- 
cessful men who make discoveries—Loeb was one of them—approach a set 
task with grotesquely simplified initial conceptualizations. Their profes- 
sional success appears to have justified them; and yet, after the work was 
completed, their continued rigid adherence to the intolerable simplism of 
the working hypothesis had its effect by bringing discredit, in part com- 
pletely undeserved, that has endured to this day. Anyone who, before 
beginning a project, delineates all the difficulties that face him, will perhaps 
never hazard a beginning. But once success is there, it will then be examined 
from a more distant vantage point and be incorporated within broader con- 
texts; then the retention of a simplism that was initially a virtue turns itself 
into a vice. 

A moderately critical simplism is permissible particularly when, as will 
be discussed in One/II/10, the research concerns a "component that is 
relatively independent of an entirety." 

In another passage Koehler says: 

An aim of the foregoing presentation was to point out examples of the kind 
of damage that can accrue for us, as biologists, in our own discipline, if we 
persist in outmoded, piecemeal thinking and forget the whole; an additional 
aim was to renew our conscious recognition of how extensive, to date, the 
group of biologists already is who sense acutely the necessity to convert 
from a purely static view to a consideration of dynamic processes and to 
become aware of the systematic character of all living things [emphasis mine]. 

If some authors preface their confessions of faith in entireties with an apol- 
ogy, this is because of a clear recognition that talking about wholes is not 
enough; what really counts is the day to day work which is guided by con- 
tinual cogitation that envelops, with the same energy, the particular and the 
most particular aspects of the questions just now being considered, as well 
as their order of proper integration within a question complex at a higher 
level of abstraction. 

H. J. Jordán (1929) in his general work on the comparative physiology of 


II. The Methodology of Biology and Particularly of Ethology 

animáis, formulated the same principie using the words, "much analysis, 
modest synthesis, and as objective, life as a whole." 

2. The Sequence of Cognitive Steps Dictated by the 
Character of Systems 

The definition of an organic entirety given by Otto Koehler States clearly 
enough that, in organic systems, there is no such thing as a unidirec- 
tional linkage between cause and effect, apart from certain exceptions 
that will be taken up later. Thus one is guilty of a fundamental meth- 
odological error if one isolates a causal relationship experimentally, or 
even conceptually, and explores this in one direction only. Yet it stands 
to reason, as Koehler says in the lines quoted above, that this is repeat- 
edly necessary; in a physiological experiment it is not possible to proceed 
in any other way than to set up some cause and to study its effect. But 
while doing this, one must remain aware that one has contrived to pro¬ 
duce an artifact. 

All sorts of approaches and all methods are permitted; only the 
sequence of their utilization is dictated by the recognition that an organ- 
ism is a system in which everything is interrelated and interacts with 
everything else. The following example of a man-made technical system 
can serve to illustrate just how obligatory is the sequence of cognitive 
steps under discussion: Let us assume that a resident of Mars has landed 
on earth and, as the first object for investigation, has found an automo- 
bile. His comprehension of this fabrication will most certainly make no 
progress before he has answered the teleonomic question, that is, before 
he has found out that this thing is an "organ" of locomotion for Homo 
sapiens. A biologist in unfamiliar territory who comes upon a creature 
that is completely new to him will also, most certainly, first seek to 
understand the species-preserving "what for" of its various organs and, 
therewith, the ecological relationships between the organism and its 
environment. In other words, he will begin his investigation using as 
broad a frame of reference as possible. 

After he has become fairly well oriented to the teleonomy of the most 
general ecological relationships, the researcher will turn his attention to 
an inventory of the parts and attempt to understand the interrelationship 
of each of them to all the others and the way they aífect one another 
reciprocally and, thereby, the entirety. One cannot master set research 
tasks if one makes a single part the focus of interest. One must, rather, 
continuously dart from one part to another—in a way that appears 
extremely flighty and unscientific to some thinkers who place valué on 
strictly logical sequences—and one's knowledge of each of the parts 
must advanee at the same pace. 

2. The Sequence of Cognitive Steps Dictated by the Character of Systems 39 

It is in the nature of our language and our logical thinking that we can 
give and receive information only in linear temporal successions. 
Goethe, who discerned this, said: "Doch red' ich in die Lüfte, denn das 
Wort bemüht sich nur umsonst, Gestalten schópf'risch aufzubau'n." 
("But I only talk into the air, for the word struggles in vain to construct 
forms creatively.") In teaching, the assimilation of information has to 
struggle with the same difficulties as in research. For this reason a didac- 
tic example can illustrate both, and again I choose one that is mechanical 
in order to elude the suspicion of endowing biological entirety with mys- 
tical properties. 

Imagine that you have to explain to someone who knows nothing 
about the thing, the functional operation of a complex machine, for 
example, an ordinary auto motor, the Ottomotor. In such a case one usu- 
ally begins with a description of the structures and functions of the 
crankshaft and the pistón, if only because these can be depicted through 
the use of graphic gestures. Then one explains how the descending pis¬ 
tón sucks a "mixture" from the "carburetor," using this terminology all 
the while, although the listener has no clue to and cannot yet imagine 
what these words mean ñor what these things might be. One hopes that, 
during this explanatory process, the listener is forming for himself a kind 
of conceptual diagram in which he holds open some empty spaces for 
"carburetor" and "mixture," spaces that will be filled in later as the con- 
cepts take on a firmer form. In principie, this is the same as the designing 
of a so-called flowchart that, with its so-called "black boxes," anticipates 
functions of a mechanism which remain, for the time being, 

The provisional sketching of the entire system is indispensable 
because the learner-listener, exactly as the researcher, can understand the 
single part, the "subsystem," only when he has also understood all the 
other parts. From what source, for example, the pistón gets the energy 
that enables it to develop a capacity for suction can first be compre- 
hended by the learner after he has understood the functions of all the 
other parts which provide the flywheel with kinetic energy; he must 
know why and to what end the camshaft runs half so fast as the crank¬ 
shaft and how it opens the intake valve on the intake stroke and holds 
the exhaust valve shut, how it closes both valves on the compression and 
explosión strokes and opens only the exhaust valve on the exhaust 
stroke; he must understand how it is that the mixture is ignited at the 
right moment of the compression stroke, and so forth and so on. Systemic 
function can be defined quite simply in that its partial functions can be 
understood all together and synchronously or not at all. 

Before at least an approximation of this synchronous understanding is 
achieved, it is senseless to turn one's attention to a more meticulous anal- 
ysis of a partial function. Measurements are also meaningless before 
one's knowledge has arrived at this stage. Quantification can contribute 


II. The Methodology of Biology and Particularly of Ethology 

nothing to reaching it, no matter how important this becomes later for 
verification and for the confirmation of what other cognitive capacities 
brought to light. 

Rupprecht Matthaei (1929), in his book Das Gestaltproblem, has com¬ 
pared the approach of the researcher endeavoring to analyze an entirety 
with that of a painter: 

A cursory slapdash sketch of the whole is gradually filled in and elaborated, 
whereby the painter develops, as far as this is possible, all parts of the pie- 
ture simultaneously; at every stage through which it is brought, the can vas 
looks as if it were finished—until the painting finally appears in its wholly 
visible self-evident completeness. 

Then Matthaei adds the following remarkable sentence: 

Perhaps, at the end, one will say that such a step-by-step progression must 
certainly have been preceded by a kind of analysis; then, though, it would 
have been one that had been guided by gestalt insight! 

This progression in a direction from the entirety of the system to its parts 
is, in biology, obligatory. 

Naturally a researcher is free to make any part of an organic entirety 
the object of his investigation; it is equally legitimate and equally accu- 
rate to examine the whole organic system within the context of its envi- 
ronmental niche, as the ecologist does, or to concéntrate interest on 
molecular processes, as the biogeneticist does. But an insight into the net- 
work of the system must always be present so that the specialized 
researcher can be oriented with respect to where, in the total organic 
structure, the subsystem he is studying has its place. Only the sequence 
of the research steps is prescribed. 

3. The Cognitive Capacity of Perception 

Wolfgang Metzger (1953) once said: "There are people who, through the- 
oretical considerations of cognition, have incurably crippled the utiliza¬ 
ron of their senses for the purpose of natural scientific understanding," 
and meant, by this, to describe particular philosophers. Paradoxically, 
however, this sentence also applies to many others, among them 
extremely acute natural scientists who believe that they proceed espe- 
cially "objectively" and "exactly" in that they exelude, as far as possible, 
their own perceptions from their research methodology. The epistemo- 
logical inconsistency of this approach is identifiable in that perception is 
granted scientific legitimacy when it serves enumeration or the reading 
of a calibrated instrument, but is disallowed when it serves the direct 
observation of a natural process. 

3. The Cognitive Capacity of Perception 


Absolutely all of our knowledge about the reality surrounding us is 
based on the reporting done by a wonderful and already well researched 
sensorial and neural apparatus that forms perceptions from data supplied 
by the sense organs. Without this apparatus, but above all without the 
objectifying capacity of its constancy mechanisms, to which gestalt per¬ 
ception also belongs, we would know nothing about the existence of 
those natural units which we cali objects—living beings, places, things. 
The messages of this apparatus, which every normal person takes on 
trust as "true," are based on processes which, although completely inac- 
cessible to introspection and rational control, are, nevertheless, through 
their functions, analogous to such rational operations as the reaching of 
conclusions and the making of computations. This, as is known, led 
Helmholtz (1887) to equate these two kinds of processes. Egon Brunswick 
(1957), one of the most successful researchers in the field of perception, 
designated what Helmholtz called unconscious conclusions as ratio- 
morphic functions and therewith expressed not only the strictly func¬ 
ional analogy but also the physiological differentiation of these two 
kinds of cognitive processes. 

The complexity of the operations accomplished by ratiomorphic func¬ 
tions is well-nigh unlimited, even among creatures whose rational capac¬ 
ites are limited to the simplest thought processes. Try to imagine, for 
example, the level of complications that the stereometric and mathemat- 
ical "computations" must attain so that an irregularly shaped object 
which revolves before one's eyes can appear to have a form that remains 
constant, that is, so that all the innumerable changes which its retinal 
image undergoes in the process of turning, can be interpreted as move- 
ments of the solid object in space and not as changes in its form. Yet the 
perception of every higher vertébrate performs this function. 

Over and above the complexity of its functions, the most striking 
attribute of gestalt perception is its specific weakness—the ease with 
which it can be misled to fallacious "conclusions" by a deceptive set of 
data. The manifold optical illusions grant us insight into the "conduit of 
computation" through which the mechanism of perception comes up 
with a finding that, in as far as its "conclusions" have been based on false 
premises, is as demonstrably false as it is incorrigible through rational 
reflections. The so-called simultaneous contrast (see below) can serve as 
an example of such a false report. 

The teleonomic function of our color perception is to make objects rec- 
ognizable by their inherent reflectional qualities. As Karl von Frisch 
(1960) has demonstrated, the bee has evolved a completely analogous 
apparatus. Bees, like humans, are not interested in ascertaining the exact 
wavelengths of light. What they must be able to do is recognize a bio- 
logically relevant object by means of its reflectional qualities. These are 
the qualities that we say, simply, are "its" color. This recognition func¬ 
tion is performed by an apparatus which separates, quite arbitrarily, the 


II. The Methodology of Biology and Particularly of Ethology 

continuum of wavelengths into a discontinuum of the so-called spectral 
colors, for no other reason than to shunt their signáis in such a way that 
the colors stand out in pairs, as so-called complementary colors, and 
through these particular combined relationships, convey the color 
white—which has been "invented" expressly for this purpose. The color 
"white" is a qualitatively consistent form of experience which does not 
correspond at all to anything simple in extra-subjective reality. Since no 
paired correspondent, in the form of actual wavelengths, exists for the 
middle of the spectrum, there to be used for compensative cancellation, 
that is, for the formation of "white," a complimentary color that we cali 
purple has been "invented" in the same way as white, whereby the band 
of colors is completed and closed to form a color ring. 

The species-preserving function of this entire apparatus lies exclu- 
sively in its capacity to compénsate for incidental variations in the color 
of the lighting and, by doing this, to abstract as constants the inherent 
reflective qualities of objects. This happens in the following way: the 
impingement of light of a particular wavelength on a part of the retina 
induces the entire remaining area of the retina not only to increase its 
readiness to receive the complementary color but also actively to create 
that color. Erich von Holst (1969,1970) demonstrated this experimentally 
by allowing a portion of the field of visión to perceive a spectral color, 
blue for example, but then beamed to the other remaining part of the 
field a "white" which did not comprise the wavelengths of the comple¬ 
mentary color, yellow, but was generated through a combination of red 
and green. Nevertheless, the subject in the experiment saw the región 
bordering the blue area as yellow and not as white. 

If in a given illumination a particular wavelength predominates, this 
will be compensated for completely by the mechanism just described, 
and that is why we always perceive a piece of "white" paper as white, 
although it may be viewed in the yellow of an electric light, in the red- 
dish light of a sunset, or in the bluish light of a foggy day. But this com- 
pensating apparatus can also easily be "taken for a ride" if the prevalence 
of a color in the field of visión is not contingent on the lighting, but is 
dominant simply because a majority of the objects being seen reflect the 
color in question more readily than others. In such a case, we see all the 
other objects that do not do this as manifestly reflecting the complemen¬ 
tary color. This phenomenon is called the simultaneous contrast. 

This phenomenon can also be made understandable through a trans- 
lation of the ratiomorphic processes into rational ones. The mechanism 
of color constancy proceeds from an established premise which 
"assumes" that the objects simultaneously present in the field of visión 
reflect, on average, all spectral colors uniformly. The apparatus does not 
"reckon" with the hardly probable possibility that, for some reason, the 
field of visión might be filled with nothing but objects which "are red," 
that is, with objects reflecting much more red than other colors, and thus 

3. The Cognitive Capacity of Perception 


"condueles," consequentially but incorrectly, that if red dominates the 
field of visión, this must be due to the lighting. But in red lighting, 
everything that reflects a "colorless" light must logically reflect green 
more intensely, or else they would at least have to appear lightly tinged 
with pink. And for the same reason, one would also see everything gray 
tinged in "contrasting" green. 

As this example (intentionally chosen as the simplest possible) shows, 
perception is in a position to process a vast amount of incoming data and 
to draw from these a completely consistent conclusión. Only this conclu¬ 
sión is communicated to the consciousness! Perception, thus, is comparable to 
a cpmputer. Erich von Holst has shown that almost all of the so-called 
"optical illusions" are based on the same principie as the simultaneous 
contrast: the premises from which the ratiomorphic operation proceeds 
are counterfeited by means of an extraordinarily improbable stimulus 
configuration and in such a way that the "computation," in itself logical, 
leads to an erroneous conclusión. 

Perception invariably evaluates relationships, configurations, but 
never absolute data. Even the most unmusical person can recognize a 
fifth or a third unmistakably, but a statement about the absolute pitch 
would be beyond him. It is immensely characteristic of perception that 
it is independent of absolute data. We recognize a melody again and 
without diíficulty whether it is played in the bass clef or in the treble 
clef. Christian von Ehrenfels (1890), one of the pioneers of Gestalt psy- 
chology, incorporated this capacity for transposition as an essential crite- 
rion of gestalt perception. 

If a comprehension of modern computers can convey more than a 
mere conceptual model for any portion of the physiology of the central 
nervous system, then it is in the región of that mechanism which, from 
the multifariousness of the sense data, extraets the biologically relevant, 
teleonomic perceptions. Far from making the impression of being, in 
principie, inaccessible to research and conducive to misleading mystical- 
vitalistic interpretations, its function—and still more its extremely 
instructive malfunctions—bear so clearly the character of the mechanical 
or, better said, the physical that, more than any of the other complex 
phenomena of life, it fortifies our research optimism. 

In spite of this, many modern researchers, and especially those who 
claim a monopoly for quantification as the only legitimate cognitive 
function, do not accept perception as a source of knowledge. In part this 
is because the messages of gestalt perception, obviously originating 
somewhere other than through the results of the rational and, for that 
reason regarded as "intuition," "inspiration," "revelation," and such like, 
are suspect to the scientist. To this must also be added the circumstance 
that those people who are extraordinarily gifted for the perception of 
complex configurations usually lack the keenness for analytical thinking. 
Only a few of the very great geniuses, but not all of them, had both. 


II. The Methodology of Biology and Particularly of Ethology 

Goethe himself considered the revelation of perception the oniy substan- 
tiai source of knowledge: "Geheimnisvoll am lichten Tag, Lásst sich 
Natur des Schieiers nicht berauben, Und was sie deinem Geist nicht 
offenbaren mag. Das zwingst du ihr nicht ab mit Hebeln und mit Schrau- 
ben." ("Nature does not allow herself to be robbed of her veil of mystery 
in the bright light of day, and what she chooses not to disclose to your 
mind you cannot forcé from her with levers and jackscrews.") The 
obvious fact that gestalt perception and rational thinking both belong in 
like manner to the cognitive apparatus of humans and are capable of 
functioning fully only together, is as difficult for some people to com- 
prehend as is the complementarity of the goal-directed, that is to say the 
teleonomic, and the causal considerations of a living system. 

Gestalt perception is the function of a computing apparatus which, in 
complexity and capacity, surpasses by far every man-made machine. Its 
great strength is to be found in the simply prodigious amount of isolated 
data it takes in, the innumerable interrelationships it registers among 
them, and its ability to abstract from these relationships the inherent 
lawfulnesses. In this it surpasses even the most recently marketed com- 
puters which are able to extract, from a superposition of a great many 
curves, the single lawfulness prevailing in all of them. 

The second, and perhaps even more intrinsic capacity through which 
perception surpasses all computers, is to be found in its ability to discover 
unforeseen lawfulnesses. Rational research and, more obviously, all com¬ 
puters can answer only those questions which have already been clearly 
formulated by means of human reasoning; at least the supposition that 
a lawfulness exists is always necessary before it is possible to provide an 
experimental or a quantifying verification. But even to those researchers 
who would prefer not to have it so, it is their own gestalt perceptions 
that will have intimated that supposition. 

The third great strength of gestalt perception, and what really makes 
it rightly the basis of all knowledge about complex systems, is to be 
found in its extraordinarily protracted retentive memory. The fact which 
Goethe so clearly recognized, that the linear succession of words is inca- 
pable of depicting, comprehensibly, complex systemic wholes, is based 
quite simply on the realization that our rational recall has long forgotten 
the facts ingested at the beginning of an explanation as it continúes to 
receive subsequent information about the same system. Most important, 
it never succeeds in grasping the relationships, going in all directions, 
that exist among the isolated data. One has only to read, for example, the 
description in an ornithological textbook of a cryptically speckled bird, 
somewhat similar to a skylark, and one will discover that one cannot 
make a mental "picture" from the descriptive phrases because one has 
long forgotten what was written about the brown speckling by the time 
one is reading the delineation of an adjacent región of the bird's body. 
That this difficulty is actually one of memory and that, in principie, it is 

3. The Cognitive Capacity of Perception 


perfectly possible to construct a gestalt from a temporal sequence of iso- 
lated data, is confirmed by televisión, a médium through which the 
transmission of facts follows so rapidly that the afterimage on the retina 
takes over those tasks on which, through the vocal transmission, our ver¬ 
bal recall has long since foundered. 

It borders on the miraculous the way in which gestalt perception can 
abstract configurations of distinctive features from a chaotic background 
of accidental stimulus data, and then retain these over the years. An 
example of this, described here, is one that will be familiar to every older 
physician. Several years ago, let us say, a certain complex of symptoms 
was observed a few times, or it could have been but once, without your 
having realized at the time that a particular gestalt quality had been per- 
ceived. But, seeing the same complex of symptoms again much later, it 
can happen that, in a flash, from the depth of the unconscious, the gestalt 
perception with the indubitably correct message is there: "You have seen 
this exact same pattern of pathology once before." 

Perception's most important characteristic for Science in general, but 
especially for comparative studies of behavior, is its capacity for storing 
configurations of data almost indeíinitely. And all those functions of 
information collection from a disordered "background," a background 
that is made up of what is to be collected as well as a lot of insigniíicant 
data, require a repetition of the extraction process. The louder the 
"noise" is on the band tuned for receiving, that much more often must 
the message be sent. "Redundancy of information compensates for the 
noisiness of channel" is the single sentence my friend, Edward Grey 
Walter, chose to summarize a lecture I once gave on the necessity of 
repeated observations. 

One day, after a long period of unconscious data accumulation, the 
gestalt that has been sought is there, often coming completely unexpect- 
edly and like a revelation, but full of the power to convince. The enor- 
mous fund of facts that the perceptual mechanism must amass before it 
is brought to a position to be able to convey such a result plays, in the 
function of ratiomorphic abstraction, a role which is analogous to that of 
collecting the basis in inductive research. And getting it to that stage of 
completion and production, perception takes just as long a time, if not 
longer, than rational induction. This explains why the discoveries made 
by a great biologist who continúes to use the same research object are 
sometimes separated from one another by decades. Karl von Frisch pub- 
lished his íirst íindings on bees in 1913; in 1920 he wrote for the íirst 
time about their ability to communicate by means of dances; in 1940 he 
discovered the mechanism for orientation which responds to the posi¬ 
tion of the sun and which has, as a prerequisite, an "internal chronom- 
eter" and also the directional instructions for returning to the hive, 
which employ a transposition of the sun's location and which, in these 
dances, are communicated by means of symbolizing the direction of the 


II. The Methodology of Biology and Particularly of Ethology 

sun through the perpendicular. In 1949 he discovered the amazing "com- 
puting apparatus" that is capable of calculating the position of the sun 
by using the plañe of polarization of the light coming from a clear sky. 

No matter how much truly diligent experimentation and conscientious 
verification also went into every one of the magnificent discoveries made 
by this great naturalist, it is certainly no accident that a major portion of 
each of them was made when Karl von Frisch was on holiday and while 
he was observing the bees of his own hives near his summerhouse. For 
one of the very pleasing peculiarities of gestalt perception is this, that it 
collects and stores information most zestfully just when the perceiver, 
immersed in the beauty of what is he observing, presumes he is pursuing 
nothing but the deepest spiritual repose. 

4. So-Called Amateurism 

It is apparent from the described qualities of gestalt that a tremendous 
amount of data must be fed into its computing apparatus if it is to fulfill 
its essential function. The amount of such data needed increases in pro- 
portion to the complexity of the lawfulness being sought and also in pro- 
portion to that part of the total incoming data that is accidental and irrel- 
evant. Among the most complex lawfulnesses we actually know are those 
which control those capacities of the sensory organs and the nervous Sys¬ 
tems of higher animáis, whose function is their behavior. Many of the 
behavior patterns which the researcher absolutely must be able to rec- 
ognize—if his investigation of the total behavior of an animal species is 
to have any success—are the exceedingly complex temporal configura- 
tions (Erich von Holst speaks of "impulse melodies") and are, in most 
cases, overlaid by still other kinds of movements which are apt to conceal 
their gestalt. One has to have seen these sequences again and again 
before their gestalt, through a sudden act of recognition, emerges from 
the background of the accidental and presents itself, as an unmistakable 
unity, to the consciousness of the investigator. 

This sudden emergence, one would like to say "eruption" of the gestalt 
is, for the observer, a qualitatively unmistakable experience which is 
always accompanied by a kind of astonishment that something so 
obvious had not been seen much earlier. Here is an example of this 
experience. At the beginning of the nineeteen-forties, when studying 
courtship movements of the Anatidae, I suddenly noticed a movement 
pattern performed by a garganey drake (Querquedula querquedulá) that 
consisted of the bird lifting back its wing at the shoulder joint, stretching 
its bilí toward the wing and then running its bilí on the inner side, along 
the secondary quills, like a small boy with an outstretched stick moving 
along the slats of a lath fence, producing a very audible, comparable 
sound. When I perceived this behavior pattern (certainly innately coor- 

5. Observing Animáis in the Wild and in Captivity 


dinated) in the garganey duck, I had a vague inkling that I had already 
seen the same thing in some other species of duck, but I could not put 
my finger on which one. Then I believed it to be the Madagascar duck 
(Anas melleri) because that was the next species in which I discovered this 
movement. Soon after this I was torced to confirm that the drakes of all 
the surface-feeding duck species to which I had access at the time exhib- 
ited the same movement pattern. 

As will be shown in a subsequent section, the instinctive movements 
or fixed action patterns, as relatively entirety-independent components, 
constitute a framework, in a certain sense, of the behavior patterns of an 
animal species. Taking an inventory of the parts or subsystems integrated 
into an entirety (already brought up and discussed in One/II/3) is the 
first and indispensable step in the analysis of a complex organic system. 
Gestalt perception is the only cognitive capacity we possess by means of 
which we can take this step. 

A simply prodigious amount of time, spent in presuppositionless 
observation, is necessary in order to collect and store the factual material 
which the great computing apparatus needs in order to be able to lift the 
gestalt from its background. Even a Tibetan priest schooled in the prac- 
tice of patience would not be able to remain stationary in front of an 
aquarium or adjacent to a duck pond or even in a blind constructed for 
observations in the open as long as is necessary to accumulate the data 
base for the perceiving apparatus. Such sustained endeavours can be 
accomplished only by those men whose gaze, through a wholly irra- 
tional delight in the beauty of the object, stays riveted to it. Here the 
great scientific valué of so-called amateurism becomes manifest: the great 
pioneers of ethology. Charles Otis Whitman and Oskar Heinroth, were 
"lovers" of the objects of their studies, and it is no accident that so many 
of the fundamental discoveries in ethology were made within the zoo- 
logical class of birds. It is one of the greatest fallacies to think that the 
expression scientia amabilis has a derogatory connotation concerning the 
branch of knowledge being referred to. 

5. Observing Animáis in the Wild and in Captivity 

The love of animáis, which constitutes a prerequisite for sufficiently sus¬ 
tained observation, can be traced to two motivational sources in man— 
to that of the hunter and to that of the herder. Among a great many 
successful ethologists, the choice of the object of their studies and their 
methodologies are determined to a great extent by the joy of stalking and 
lying in wait for animáis, in short, by the pleasure of "outwitting" them. 
Studying a captive animal is hardly any fun at all to them and, in fact 
this seems to some of them to be "unsportsmanly." They easily rational- 
ize this purely emotional antipathy through the use of the argument that. 


II. The Methodology of Biology and Particularly of Ethoiogy 

with captive animáis, one can never know for certain how much of their 
behavior has been altered by the abnormal conditions of their confine- 
ment. And in point of fact, in order to eliminate this source of error, 
repeated observations of animáis in a natural habitat are absolutely nec- 
essary. But the primary advantage of observations made in a natural set- 
ting is that they can comprehend directly the ecological integration of 
the animal species being studied. 

The "hunter type" of behavior researcher, embodied in my friend, 
Nikolaas Tinbergen, is the opposite of the "herder type" who wants to 
own animáis and, above all, to breed them and increase their number. 
Herders stalk animáis only in order to capture them and, subsequently, 
to keep them. C. O. Whitman and Oskar Heinroth were such animal 
keepers; I, myself, am also one. The single great disadvantage associated 
with observing the behavior of captive animáis has been stated above. 
Yet, at the same time, there are many advantages which compénsate for 

The most important of these advantages is the opportunity provided 
the researcher who keeps animáis to observe the behavior of several spe¬ 
cies simultaneously and in juxtaposition. This advantage is even further 
enhanced by the tendency amateurs often have of limiting themselves to 
particular classes of animáis; in many cases they begin with a small 
group, a genus, or a family, and then only gradually do they expand their 
interests to inelude larger taxonomic categories. My first object of study 
was the mallard; then my interest slowly extended to embrace the entire 
group of Anatidae. Heinroth and Whitman were also lovers of animal 
groups: from early youth onward Whitman kept and studied pigeons; 
Heinroth kept and studied Anatidae. Discovering the homology of those 
innate sequences of coordinated movements which we cali fixed motor 
patterns, which proved to be basic for all comparative studies of behav¬ 
ior, could have been achieved by no one except a lover of animáis, and 
particularly a group of animáis—in other words, by a so-called amateur. 

Another great advantage for observing animáis in captivity is to be 
found, paradoxically, among just those disturbances of behavior which 
are brought about by the unnatural conditions of the environment. If, 
for example, a researcher is able to observe the way in which a wolf in 
the wild carries the remains of a kill to a covert place, digs a hole in the 
earth there, pushes the piece of plunder in with his nose, shovels the 
excavated earth back—still using his nose—and then levels the site 
through shoves with the nose, the teleonomic question concerning this 
behavior sequence is easy to answer, but the question concerning the 
causal origin of this behavior pattern remains completely unanswered. 
If, in contrast to this, the observer of captive animáis sees the way in 
which a young wolf or dog carries a bone to behind the dining room 
drapes, lays it down there, scrapes violently for a while next to the bone, 
pushes the bone with his nose to the place where all the scraping was 

5. Observing Animáis in the Wild and in Captivity 


done and then, again with his nose and now squeaking along the surface 
of the parquetry flooring, shoves the nonexistent earth back into the hole 
that has not been dug and goes away satisfied, the observer knows quite 
a lot about the phylogenetic program of the behavior pattern. If he then 
sees (what every dog owner has experienced) that the pup, after he has 
once gone through the described behavior sequence out in the open, 
never again makes the futile attempt to bury a piece of booty on a solid 
surface, then it will be clear to him that the presence of certain associated 
stimulus configurations function as a reward to "condition" the behavior 
to the normal situation and that, conversely, their absence works to 
weaken, that is, "extinguish" it. In this way the observation of perfor¬ 
mance failures caused by captivity often produces an unexpected amount 
of information about the nature and the composition of the sequences of 
behavior observed. 

Disturbances of behavior that are pathological are a second source of 
knowledge. Just as pathology in general is one of the most important 
sources of physiological knowledge, and just as the pathological defi- 
ciency of a particular function makes the researcher aware of a particular 
mechanism, just so does the student of behavior very often learn some- 
thing of the greatest importance from pathological disturbances of 

One of the most prevalent of these disturbances among captive ani¬ 
máis is the diminution of the intensity with which certain instinctive 
movements, and the complex behavior patterns they combine to form, 
are carried out. A typical example of this is the construction of nests by 
caged birds. How often nest-building activities are begun in the aviaries 
of amateurs and in zoos, and by such a great variety of birds, and then 
how seldom it is that a nest characteristic of the species is ever completed. 

This pathological disturbance of a phylogenetically programmed 
behavior pattern is very similar to the normal incompleteness which we 
observe in the ontogénesis of young animáis. In 1932 I wrote the 

Among the young [I am writing here about young birds], drives intended to 
be directed toward a particular object at first develop without this object 
being present, and then only secondarily are they conditioned to an appro- 
priate object. Among animáis having a high level of intelligence and within 
whose chains of instinctive behavior sequences many such links are inter- 
calated, that is, links which are modifiable through personal experience, the 
actual purpose of a course of action which does not achieve its real goal 
might be that of preparatory practice for the young animáis and is thus of 
biological significance. 

The behavior sequences which remain uncompleted because of a lack of 
intensity frequently stop just short of fulfilling their species-preserving 


II. The Methodology of Biology and Particularly of Ethology 

function. This fact was what first convinced me that the animal had no 
idea about the teleonomy of its own behavior and that the purposive 
psychologists were quite wrong in supposing that the survival valué of 
any action was to be equated with the goal at which the animal, as a 
subject, was striving. I also wrote in the monograph mentioned above 
that an instinctual programming of behavior must be postulated when- 
ever "there is an obvious incongruity between the normal intelligence 
capacities generally observed in the animal concerned and those which 
would have to be present for the completion of the given behavior pat- 
tern," and secondly, whenever "imperfections appear in a behavioral 
sequence which distinctly indicate that the animal itself is not conscious 
of the purpose of the behavior pattern concerned." 

Because the intensity of so much instinctive behavior is directly 
dependent on the physical condition of the animal, extreme caution 
becomes particularly necessary if, from the absence of a particular behav¬ 
ior pattern in a captive animal, a conclusión were to be reached concern- 
ing its presence or absence in the species as a whole. Concluding that a 
particular behavior pattern is not present among a certain animal species 
is, in principie, inadmissible when this is based only on observations of 
caged animáis. When Gustav Kramer (1950) demonstrated the extreme 
aggressiveness of Lacerta melisselensis, a lizard of the Muralis group, his 
findings were questioned by some herpetologists who, for years, had 
kept and even bred the same species in terraria; none of them had ever 
witnessed any serious fighting. Under normal terrarium conditions, 
under which the animáis stay alive for years, one can keep several males 
of this species together with no trouble, but in their natural habitat this 
species belongs to those few animáis among which an immediate and 
fatal outcome is observed in battles between rivals. Kramer demonstrated 
that the decrease of aggressiveness among captive animáis is caused by 
climatic conditions; these animáis need cool air for breathing while the 
blood heat necessary for normal behavior must be supplied by radiation. 
Statements concerning the non-existence of a particular behavior among 
captive animáis carry no weight at all. 

On the other hand, the positive observation of a fairly complex behav¬ 
ior pattern in a captive animal justifies a confident assertion that the 
behavior in question is characteristic of the species. In other words, the 
influences of captivity can lead, essentially, only to the disappearance of 
behavior patterns, but can never give úse to a complex and, above all, an 
obviously teleonomic behavior. When, at an ethology congress in Den 
Haag, Janet Kear-Matthews reported on her successful breeding and 
rearing of the Magpie goose (Anseranas semipalmatus) and how this bird 
was the only one among all the Anatidae to consistently feed its young 
and how the young had a repertory of appropriate begging movements, 
a well-known Australian field observer maintained, during the discus- 
sion, that the behavior of Anseranas that Kear-Matthews had observed 

6 . Observing Tame Animáis Not Kept Captive 


and had shown in a film was an artifact induced through circumstances 
of captivity. He based his statement on the fact that he had never seen 
anything like this during his observations in the field. This way of rea- 
soning bears witness to the most profound ethological ignorance. 

6. Observing Tame Animáis Not Kept Captive 

A possible compromise between observations in the field and keeping 
animáis caged can be achieved with some highly organized vertebrates 
if one rears young animáis of a social species within a suitable biotope 
and then observes these tame individuáis as they move about unhin- 
dered. The advantages of this research method are obvious. Some limi- 
tations are set by the characteristics of what is termed imprinting; these 
play an important role, particularly among birds. Among ravens, parrots, 
herons, and others, it is difficult to rear the birds so that their filial 
instinctive patterns and their social attachments do, actually, become fix- 
ated on humans, but not those behavior patterns associated with sexual- 
ity and fighting among rivals. An adult great yellow-crested cockatoo 
(Cacatue galerita) or a male raven can frústrate all further observations if 
he begins to treat the person who has reared him as a rival. Geese (Anser- 
ini) of the most diverse species are so well suited for the research method 
being discussed here because the behavior patterns directed toward the 
parent and toward the social companion can easily and enduringly 
become fixated on the human caretaker, but a sexual imprinting of these 
birds to humans is next to impossible. 

If, for the purposes of observation, one attempts to establish a popu- 
lation of non-captive animáis or tries to incorpórate within a natural pop¬ 
ularon an individual reared tame, one sometimes comes up against an 
unexpected difficulty, especially among the most highly developed 
mammals. Even if settled with success within a living space that corre- 
sponds extensively to the natural habitat, animáis reared by humans 
form a type of society that is not typical of their species. Attempts made 
by Americans to settle in spacious enclosures those chimpanzees who, as 
test animáis, had earned the right to retirement, miscarried completely. 
And Katharina Heinroth (1959) has reported graphically on her futile 
attempts to intégrate a hand-reared young female baboon (Papio sphinx) 
into a troop of animáis of the same species that lived on the large cliífs 
for apes at the Berlin Zoo. Again and again the young female reared by 
Dr. Heinroth stirred up the greatest indignation among her conspecifics 
through behavior that was not "socially acceptable." As Dr. Heinroth 
expressed it, "She made social gaífes." The discrepancies are significant 
because they permit the conclusión that traditional social norms are 

Among hand-reared wild boar (Sus scrofa), the difficulty just described 


II. The Methodology of Biology and Particularly of Ethology 

does not arise. Apparently the entire behavior inventory they need for 
social interaction is firmly set genetically and does not have to be sup- 
plemented by social tradition. This problem of social tradition can be 
properly studied only through comparisons between a natural popula- 
don of animáis and another free-ranging population reared by man. 

In the observation of highly organized creatures, the methodological 
ideal is achieved when it has become possible to accustom free-ranging 
wild animáis to the observer to such an extent that their behavior is not 
influenced by his presence and he can, in fact, conduct experiments with 
them within a natural environmental setting. This research methodology 
is of particular importance when investigations are undertaken with pri¬ 
mates—because their social tradition plays such a prominent role, so 
prominent that tame animáis which have been kept captive and then set 
free cannot be expected to exhibit social behavior characteristic of their 
species. That is why the observations of chimpanzees made by Jane 
Goodall (1971) received the recognition they deserved. The observations 
of baboons made by Washburn and de Vore (1961) and especially the 
studies done by Hans Kummer (1957, 1968, 1971, 1975) show, through 
their results, how very much this extremely time-consuming and sacri- 
fice-exacting method pays off. 

7. Knowing Animáis: A Methodological Sine Qua Non 

When Erich von Holst (1969/70) was conducting his classic experiments 
involving the electrical stimulation of the hypothalamus of chickens, he 
emphasized repeatedly that such an investigation could not forego, as a 
basis, the most exact knowledge of the entire behavior inventory of the 
species being studied. For this reason, he persuaded Erich Baeumer 
(1955) to join him as a co-worker. Baeumer, by profession a country doc¬ 
tor, had through his love for these animáis accumulated a profound 
knowledge of the behavior of all the breeds—an indispensable resource 
for von Holst's experiments—and could provide an exact diagnosis of a 
particular behavior pattern even when presented only with its most min¬ 
imal indication or with just a pattern fragment. In his time and for the 
same reason, W. R. Hess (1954) needed Paul Leyhausen to join him for 
the evaluation and interpretation of films made of the hypothalamus 
stimulation experiments done with cats. Leyhausen had the entire 
behavior inventory of cats "at his fingertips," an inventory that is much 
more complex than that of the domesticated chicken. It is my opinión 
that, without this prerequisite knowledge, hypothalamus stimulation 
experiments are meaningless. 

I would like to propose the speculation that this prerequisite pertains 
to all investigations of animal behavior. Even if one consciously limits 
the research interest to a very small subsystem of an entirety, it seems to 

8 . The Non-Obtrusive Experiment 


me that knowledge of the total behavior inventory of the studied animal 
species is absolutely indispensable. Without this knowledge the 
researcher cannot possibly know what answers the system is giving 
through its responses to his own operations, whatever these might be. 

If only in order to know whether an observed behavior pattern cor- 
responds to the proper behavior of the species in its natural setting, or if 
the behavior is pathological, the researcher must know his animal species 
very thoroughly. A great number of errors can be attributed to the cir- 
cumstance that a pathological loss of intensity or the complete lack of 
certain centrally coordinated movements were regarded as "normal." For 
use as a research model and as one part of the knowledge of animáis 
must also be included a certain measure of what, among older physicians, 
is called the "clinical eye." This is something that can be acquired only 
through extensive experience. I regret that a very large proportion of the 
younger researchers who consider themselves ethologists show a deplor¬ 
able lack of knowledge of animáis and particularly a lack of the "clinical 

8. The Non-Obtrusive Experiment 

All that has been said in Section 2 of this chapter about the systemic 
character of the organism and the sequence of operations prescribed 
implies a series of important consequences of which the experimenter 
must be aware and to which he must conform. 

It stands to reason that any encroachment by experimentation makes 
sense only after the system has been analyzed far enough to give some 
notion of the subsystems comprising it and the way in which they inter- 
act. These provisional notions must possess a sufficient probability of 
being correct in order to make their testing worthwhile. Hazarding just 
any action, "just to see what will happen," has but minimal prospects for 
any acquisition of knowledge. 

At any rate, the experimenter must limit his encroachments to mea- 
sures that are sure not to upset the equilibrium of the entire system. The 
smaller his experimental interference, the more certain he can be that 
the response shown by the whole organism is indeed caused by the 
experimental influence. 

Experiments fulfilling all the above postulates have been termed, by 
Eckhard Hess (1973), "unobtrusive" experiments. 

If experimental questions are put too early, this can easily lead to an 
underestimation of the complexity of the system being examined. Con- 
sideration of the system as an entirety and questions concerning the 
teleonomy of its individual parts are mandatory. However, they may not 
be elevated to a goal in themselves. During the preparatory work, the 
goal of the research must be kept in mind and this goal is always the 


II. The Methodology of Biology and Particularly of Ethology 

causal analysis of the entire system. The necessary cooperation of gestalt 
perception comprehending the entirety, on the one hand, and the anal¬ 
ysis and the experimentation, on the other, has been depicted by Otto 
Koehler (1933) in a beautiful comparison. 

If our ignorance of the essence of life can be likened to a large mountain, 
then the individual branches of biology and their auxiliary Sciences are tun- 
nels that penétrate into its interior from the most varied points of departure, 
extending in various directions, some into the depths, others nearer the sur- 
face. Researchers working with the methods of natural Science are the tun- 
nel crew; one crew scrapes the wall, another blasts away at the tunnel's end, 
a third removes the rubble. But when a great event occurs, such as Boveri's 
and Sutton's hypothesis that the processes in the chromosomes and those of 
heredity can be bracketed to one explanation, so that from both sides pre- 
dictions not only become possible but are actually fulfilled, then this intel¬ 
ectual achievement is comparable to the final blow of the pickax that breaks 
through the remnant of wall separating two tunnels meeting within the 
mountain's interior. 

As a contrast to an undertaking of the tunnel work that has been described 
here, what most readily comes to mind is a walk along the trails on the sur- 
face of the mountain. After their week of daily work, the men will certainly 
be granted an unlimited panoramic view on Sunday. Eyes that have become 
accustomed to the dark can delight in the brightness of the day; and besides 
the respite that is afforded as a result of the week's work, direct benefits can 
also accrue from such Circuit tours if they serve to confirm the direction of 
the tunnels, to observe the progress made by a neighbor and, together with 
him, to plan future tunneling. Such circuits of the surface become of ques- 
tionable valué only when they are repeated too often and become an end in 
themselves, or when they lead to a disregard of the efforts being made by 
the men in the mountain's interior. No number of treading hikers will ever 
trample the mountain down; it would appear much more likely that the 
mountain could collapse if new tunnels continué to be bored through it and 
stronger and still stronger detonations are ignited. But this eventuality is 
part of the far distant future, for the mountain is immense. 

Every researcher has every right to choose where, within a system, he 
or she wants to bore in depth—as long as the researcher knows where 
that place is located with reference to the entire system. It is just as legit¬ 
ímate to begin research at the highest level of integration of the organic 
system, that is, at the level of its interaction with the immediate environ- 
ment, at the ecological level, as it is to choose a small subsystem to be the 
object of study—provided the prerequisite specified above is fulfilled. 
The only strategy that is dictated to us by the very nature of the system 
is, as has already been specified, the sequence in which the survey of the 
entirety and the analytical experiment are initiated: the latter must fol- 
low the former. 

What must constantly be kept in mind are the various side effects on 

8 . The Non-Obtrusive Experiment 


the entire system and, therewith, the repercussions on the investigated 
subsystem any experimental encroachment might have. This problem is 
that much easier to solve the more the examined part or subsystem bears 
the characteristic of a relatively independent component of the entirety 
(One/II/lO). In addition to this, one must always take into consideration 
that an experimentally established conditional change can have a similar 
and predictable effect a second time only if the entire system finds itself 
at the time of both influences in exactly the same condition. H. J. Jordán 
(1929) has made this imaginably concrete by means of a very neat 
mechanical example. If, within a large railroad yard, switch X is moved 
from its connection with a sidetrack so that the main track is thrown 
open, express train Y will collide with freight train Z. The reproducibil- 
ity of this successful maneuver depends, understandably, upon all the 
switches in the system being set as they were in the first experiment and, 
moreover, that the trains Y and Z are in the exact same places at the 
moment of the shunting, moving in the same directions as before and at 
the same speeds. Transferring switch X from the siding at any other than 
the prior time would result in a completely different development. 

If the system being examined is a relatively small subsystemic part of 
the organic entirety, it is in some cases possible to achieve by artificial 
means the sameness of prevailing circumstances which, in Jordan's illus- 
tration, is represented by the necessity of prior sameness for all the 
switch settings and train movements. In physiology this method is gen- 
erally applied and is tolerably successful. But the system of sensory and 
nervous functions underlying the behavior of higher animáis is quite the 
most complex system that we know. It is naive to imagine that one can 
create, for an intact organism, a repetition of completely identical cir¬ 
cumstances and processes by simply confining it to "constant and strictly 
controlled laboratory conditions." 

As will be discussed in Two/III/2 and 3, concerned with readiness- 
increasing and releasing stimuli, a higher animal requires a vast number 
of continuously effective, often quite complicated stimulus impinge- 
ments in order to be able to maintain its good health and not show signs 
of pathologic behavior. The "controlled laboratory conditions" inevita- 
bly cancel out an unpredictable number of stimuli situations indespens- 
able to the animal, at the same time producing an equally unpredictable 
number of chaotic, abnormal stimuli. In her classic work on the spon- 
taneity of aggressive behavior in the coral-reef fish, Microspathodon chry- 
surus (Pomacentridae), Ann Rasa (1971) showed experimentally how, 
even for such a "low" vertébrate, a monotonously maintained environ- 
ment predictably produces a pathological waning of over-all excitability. 

Anyone who has kept higher animáis knows all too well how diíficult 
it is to avoid the consequences of the monotony of conditions under 
which the captive animal has to live. Even among free-flying greylags it 
has been demonstrated that a measurable reduction of general excitabil- 


II. The Methodology of Biology and Particularly of Ethology 

ity is brought about if the birds live constantly on one and the same pond 
or small lake. Compared to the wild geese presently living (one could 
hardly say "kept captive") at Grünau in the Alm valley, which fly a dis- 
tance of 16 km daily from where they sleep to where they they feed and 
back again, and which nest at still other locations, our Seewiesen geese 
now appear to have been somewhat sleepy. At Seewiesen, rival ganders 
almost never demonstrated the aerial combat behavior that, in early 
spring, is a daily occurrence at Grünau. 

In view of the unpredictable alterations in the behavior of higher 
animáis that are caused by even the mildest forms of semi-captivity, all 
attempts to achieve uniformity of response to specific stimulation by 
"controlling" environmental conditions would appear to be hopeless. 
Recognizing this, some research workers, and particularly those who are 
profoundly conscious of the systemic character of their study object, have 
become completely averse to experimentation generally. 

If one looks through the most modern literature concerned with 
behavior research, one could almost get the impression that the attributes 
"system conscious" and "experimental" contradict one another and can 
hardly ever be applied simultaneously to the same investigation. Alto- 
gether too many experiment oriented behavior researchers working 
exclusively in laboratories—as a majority of American psychologists 
do—are experimentally highly gifted but are often also afflicted with a 
complete lack of biological and ecological sense while, at the same time, 
some behavior researchers who are well able to think in biological and 
ecological terms, dedicate themselves only to observations in the field 
and choose not to know very much about experiments, particularly those 
experiments carried out under laboratory conditions. 

In view of this deplorable State of affairs, all modern students of behav¬ 
ior might take to heart the following statement written by the botanist, 
Fritz Knoll in 1926: 

In order to make an experiment with animáis useful to floral ecology, the 
experimenter must first acquire a thorough knowledge about the general 
way of life of the animal that is to be studied. This can be accomplished only 
through prolonged periods of cióse observation in the natural environment 
of flora and fauna associated with it. Only after such preparation should one 
proceed to designing and making an experiment. Initially it is advisable to 
carry out the planned experiment within the natural habitat of the studied 
flowers and insects. Certain experiments for which the original environ¬ 
ment is not suitable will be carried out in the open, in situations more 
appropriate to the experiment's purpose. Finally, for such experiments 
which should be made under very specially adapted circumstances, one will 
have recourse to the laboratory. Appropriate laboratory experiments 
arranged with all necessary caution often permit us to ascertain the very last 
nuances about a phenomenon which, in the too complex natural environ¬ 
ment, have eluded a clear analysis. 

9. The Deprivation Experiment 


These sentences are by no means valid only for those behavior patterns 
through which the insects interrelate with the blossoms that they visit. 
They are unconditionally valid for the study of behavior as such. There 
is no such thing as a behavior, just as there is no such thing as an animal 
life form, that can be understood in any other way except within the con- 
text of the ecology of the species. The sentences quoted above are even 
valid, in fact, for pathologic behavior because, as is well known, the path- 
ologic can be defined only by having recourse to ecological concepts. 

The strategy of the experiment so clearly outlined by Fritz Knoll cor- 
responds exactly to what Eckhard Hess termed the "unobtrusive experi¬ 
ment," a designation which I, in the German-language edition of this 
book, called the "non-disruptive" experiment. What is essential and 
what also emerges from Knoll's writings is this: experimental impinge- 
ments on the natural conditions of an animal's existence should change 
those conditions as little as possible. The smaller the impingement, the 
greater is the probability that it will take effect on only a single part of 
the systemic whole and, therewith, the observed effect will actually be 
caused in a relatively direct way by the experimental impingement. The 
grand master of the unobtrusive experiment is Nikolaas Tinbergen, 
whose books and articles, particularly for the sake of his methodology, 
should be studied thoroughly by every budding ethologist. His article 
on the grayling moth (Eumenis semele L.) (1943) proves that the entirety 
oriented experiment can also, ultimately, lead to quantifiable 
verifi catión. 

9. The Deprivation Experiment 

That experiment involving the withholding of specific experience plays 
a decisive role in the unfortunate and heavily belabored discussion of 
the question concerning whether innate, that is, phylogenetically pro- 
grammed behavioral elements can be differentiated from acquired 
behavior elements. This experiment can fulfill its intended task only if 
the experimenter has learned to think in biological terms and only if he 
follows the methodological rules put forward in this chapter. 

The kinds of blatant blunders that are committed against the demands 
of the unobtrusive experiment can be illustrated by means of an example. 
Young chimpanzees were reared in complete darkness and when, 
finally, they were brought into the light and did not possess the capacity 
of image visión, it was concluded that the capacity of point discrimina- 
tion and, therewith, seeing images must be "learned". Although visual 
experience is necessary to maintain the mechanisms of innately specified 
visual processing of sensory data, and in perfecting the calibration of 
binocular depth visión, it is misleading to State that point discrimination 
must be learned. 

What, if anything, can be concluded from a deprivation experiment? 


II. The Methodology of Biology and Particularly of Ethology 

Cogent conclusions can be drawn only if, in spite of the withholding of 
very specific information, the experimental animal still unmistakably 
possesses just this information. The premises for this reasoning are sim¬ 
ple: adapted behavior always presupposes that information about the cir- 
cumstances of the environment, to which the adaptation relates, has got 
into the organism. There are only two ways this can happen: either dur- 
ing phylogeny, during which the species interacted with just these envi- 
ronmental givens, or the individual in the course of its life had occasion 
to experience that which provided the information. Phylogenetic adap¬ 
tation and individual adaptive modification of behavior are the only two 
possible ways extant for the acquisition and storage of information. What 
the deprivation experiment can do, in favorable instances, is give us an 
answer to the question: From which of these two possible sources has the 
information for behavioral adaptation come? If it is possible to exelude 
with certainty one of these sources, then we know that the other has 
supplied the information. For the reasons already given, it is easier to 
exelude with certainty individually acquired information than it is to 
exelude information acquired phylogenetically. 

If we rear a male stickleback (Gasterosteus aculeatus) isolated from its 
conspecifics from the egg stage onward and when, during the following 
spring, it turns out that the fish, in a selective and specific way, responds 
to a display of "red-below" with the movement patterns of rival fighting, 
or if a young dog which has never before dug up earth and, without any 
previous experience, performs each of the bone-burying movements in 
the correct sequence (as was described in Section II/5), we know with 
certainty that the adaptive information for these teleonomic behavior 
patterns is contained in the genome. Remarkably, this certainty is con- 
tested by some groups of scientists. It is said again and again that the 
organism could have acquired information through learning while still 
in the egg or in the uterus. One must then ask: But from where? As has 
already been said in the introduction to this book, the assumption made 
by Kuo (1932) and other behaviorists, that the mechanisms of learning 
"know" without any previous experience what is and what is not useful 
for the organism, contains the covert postulation of a prestabilized har- 
mony to which the great vitalist, Jakob von Uexküll (1921), overtly tes- 
tifies. If one does not believe in miracles—and a prestabilized harmony 
would be one such—it remains simply incomprehensible where, for 
example, within the aquarium in which the young stickleback was 
reared—among all the diversity of its animal and plant world—the 
information should be contained that the rival to be attacked is red on 
the ventral side. 

But this certainty exists only for those cases in which, despite the with¬ 
holding of experience, the information appears undiminished. Stating 
the reverse, that is, to say after conducting the experiment that the miss- 
ing information must be learned, has far less certainty. If, supposing, our 
stickleback had not produced the specific reaction to the wedding suit of 

9. The Deprivation Experiment 


a rival, we would in no way be in a position to State with certainty that 
this reaction had to be learned individually. For, in such cases, there is 
always another possible explanation. This is that, through the measures 
which were necessary in order to withhold from our experimental ani¬ 
mal certain specific information, at the same time and without intending 
to, some other "components" essential to the realization of the behavior 
fixed in the genome were also withheld. Moreover, there is yet another 
possibility. When arranging the experiment cited in our example and 
when first introducing the rival fish to our experimental stickleback, 
some unspecified stimulus combination necessary to the release of the 
behavior being studied might have been missing. Male sticklebacks as 
well as a host of other territorial animáis fight only in an area they have 
thoroughly explored and which they regard as their very own. If they 
are put into an unknown environment and together with that object 
which is expected to release in them rival fighting, they will most cer- 
tainly not react as anticipated. 

A second example of the kinds of mistakes that can be made when 
using deprivation is afforded by the experiments conducted by Riess 
(1954). He reared rats in such a way that they were never given an oppor- 
tunity to lift or to carry any sort of solid object. When he then, under 
strictly controlled and standardized experimental conditions, placed 
them in enclosures supplied with nest materials, the rats built no nests. 
From this he concluded that nest building is impossible for rats if the 
handling of solid objects has not been learned beforehand. What he over- 
looked was that the time span allowed to the experimental animal was 
much too short. In absolutely new surroundings, such as the test situa- 
tion was for the rat, the animal will spend a very long time in exploring 
the new container, mainly trying to get out. It will not even begin to 
build a nest before it has thoroughly familiarized itself with the new 
container. No rat, with or without the normal experience of managing 
solid objects, would have begun to build a nest within the time that was 
experimentally allotted. Before Eibl-Eibesfeldt (1958) initiated his own 
experiments with rats which were to prove Riess's conclusions wrong (as 
will be explained later), I had the greatest difficulty convincing Eibl that 
it was necessary to repeat the Riess experiments and, as a particular step, 
inducing him to place an oíd and experienced rat in the same test situa- 
tion that had been used by Riess. Eibl regarded all of these repetitious 
research steps as ridiculous because, through his excellent knowledge of 
animáis, he knew with certainty that a rat placed in a strange environ¬ 
ment will not build for a long time. But this was something that Riess 
did not know. 

Very often innate information also fails to appear if the experimental 
animal is not in first-class physical condition and is not capable, there- 
fore, of achieving the necessary degree of general arousal. An example 
of an error which I committed more than forty years ago is relevant in 
this context and can serve as an illustration. I reared young red-backed 


II. The Methodology of Biology and Particularly of Ethology 

shrikes (Lanius collurio L.) and observed the ontogeny of the interesting 
behavior pattern by means of which these birds impale captured prey on 
thorns for safekeeping and storage. My birds did perform normally the 
motor pattern of wiping captured prey along twigs and pressing down 
when, during this process, resistance was encountered, but they showed 
no propensity whatsoever for orienting this motor pattern to thorns on 
which the captured prey could be firmly impaled. This orientation they 
learned later when a run-off of the complete sequence of movements was 
successful, quite accidentally, several times. Nothing was easier to 
assume (at least then) that just these successes, that is, the reporting back 
of the successes—the desired result having been achieved—involved a 
conditioning effect, and that recognition of the thorn is, in this way, 
learned. At that time I spoke about an instinct-learning intercalation and 
I was as yet completely unaware that everything that is conditioned by 
reinforcement is learned in this way and that everything that is learned 
is based on a phylogenetically evolved program. 

Later experiments which U. von St. Paul and I initiated showed that 
the innate "recognition" of the thorn by red-backed shrikes (Lanius col¬ 
lurio L.), as well as by greater shrikes (Lanius excubitor L.), is completely 
innate. Young birds of these species, with no experience whatsoever, 
approached all objects resembling thorns—such as nails driven through 
a perch—at once, nibbled at them exploratively and then turned their 
attention immediately toward an object, even some substitute object (a 
piece of paper, the dried wing of a butterfly) in order to be able to impale 
this directly in a workmanlike manner and in accordance with all the 
rules of the art. Reversed, that is, when presented with a juicy and ideal 
object for impalement, such as a newly hatched chick, or just the piece 
of one, the appetence for a thorn was activated at once. St. Paul and I 
then tried to rear red-backed shrikes as badly, as negligently and inade- 
quately as I had apparently reared them forty years before. But this mis- 
carried completely; U. von St. Paul takes such good care of birds that she 
was simply incapable of bringing herself to rear them as badly as would 
have been necessary to cause the previous behavior deficiency. 

The fact that the presence but not the absence of phyletic information 
can be conclusively affirmed by means of the deprivation experiment led 
some of those of the behaviorist opposition to the erroneous belief that 
ethologists desired to disavow the existence of learning processes; these 
opponents labeled the above, absolutely irrefutable facts a "highly pro- 
tective theory," as if, on the basis of these findings, we ethologists would 
deny that learning can be ascertained. 

There are cases in which reciprocal reasoning processes permit, to a 
certain extent, the exclusión of the presence of innate information with- 
out having to conduct special experiments for confirmation of this, and 
allow the assertion that the teleonomy of a particular behavior pattern 
has been achieved through individually acquired information. This is 

9. The Deprivation Experiment 


possible when, for example, an organism masters in an unmistakably 
teleonomic way an environmental situation that its species demonstrably 
could never before have come up against. 

Curt Richter told me that he once experimented (1942-43) with var- 
ious animáis, among them three-toed sloths (Bradypodidae), in a way 
that involved a selection of foodstuífs by the test animáis. In their natural 
environment sloths eat tree leaves almost exclusively and, like many 
leaf-eaters, are not easy to please or satisfy with substitute foods when 
kept in captivity. By oífering his captive animáis just about every kind of 
fresh vegetable matter available on the open market and by letting them 
choose from this what they wanted to eat, he was able, subsequently, to 
feed each of them most successfully on the "menú" it had itself selected 
(personal communication, 1955). With rats he set up and conducted the 
following significant experiment. He presented them with the most var- 
ied foodstuífs broken down, as far as possible, into their Chemical com- 
ponents—protein, for example, in the form of its constituent amino 
acids. Each component was presented in a particular dish to an amount 
that had been previously weighed, and what was eaten was measured. 
The rats took from each of the dishes just that amount necessary to con- 
stitute a well-balanced diet. The amounts of the individual amino acids 
the rats consumed corresponded to the proportional amounts that would 
be necessary for the synthesis of protein. 

Because no animal since the appearance of life on earth can have been 
in a position to put protein together from amino acids, it is impossible 
for rats to have any inborn information about this, or about which pro- 
portions of those components of their food belong together so as to com- 
prise a healthy diet. This knowledge, thus, must have been acquired dur- 
ing an individual animal's life and the way this happens was discovered 
by John Garcia (1967) and his colleagues. It was impossible for them to 
get a rat to leave oíf eating a particular foodstuíf by infliction of pain, by 
electric shock, by forced swimming in coid water, or by the effects of 
other quite gruesome punishments; an association between these stimuli 
and the eating of a particular foodstuíf was absolutely impossible to 
establish. If, on the other hand, the punishment stimulus used was a 
small dose of x-rays or the injection of a small amount of apomorphine, 
the eífect of both being intestinal nausea, the rats at once related this 
disagreeable experience to that foodstuíf which they had last eaten, fol¬ 
lowing exactly Wilhelm Wundt's oíd law of successivity and contiguity. 
In only one single set of circumstances did their reaction not follow this 
law: when the rats were given a new foodstuíf along with an extensive 
sequence of dishes, all of which were known to the animáis, they related 
the subsequent nausea to the new food, even when this was by no means 
the last one to be eaten. The wonderful teleonomy of this innate dispos- 
tion for learning is striking. As this example shows us, even when we 
know with certainty that the teleonomy of a behavior is based on learn- 


II. The Methodology of Biology and Particularly of Ethology 

ing, we can understand this process of learning only after we compre- 
hend the phylogenetically evolved System into which the process has 
been incorporated. 

With the exception of rare cases of this kind, it is more difficult to 
prove the assertion that the teleonomy of a behavior is based on learning 
and not on innate information than it is to prove the opposite. For all 
that, a researcher who knows his animáis thoroughly and who possesses 
the indispensable "clinical eye" for their States of health can almost 
always determine, with quite reliable certainty, whether the absence of 
a particular behavior is the result of a lack of learning opportunities or 
the result of a lack of sufficient "condition" or "tone" on the part of the 
experimental animal. 

The only procedure that justifies concluding a lack of learning oppor¬ 
tunities consists of first excluding those opportunities and then elimi- 
nating the disturbances thus caused, through a deferred presentation of 
the learning opportunity. When Eibl-Eibesfeldt studied the nest-build- 
ing behavior of inexperienced female rats (1958), he reared them in 
receptacles containing no objects with which they could perform the 
"carrying-to-nest" motor patterns. Even the food they were given was 
reduced to tiny particles. Yet the first group of experimental animáis had 
to be let go because they treated their own tails as nest-building material. 
They went out from where they had slept, in search, and "found" their 
tails. They carried these to the place they had chosen for a nest site and 
carefully laid them down there. A second group of experimental animáis 
was reared after their tails had been amputated while they were still in 
the nest, and when they were not yet sexually mature they were given 
shredded blotting paper as nest material. The reaction of the experimen¬ 
tal animáis differed from that of the normal, control animáis only in its 
intensity; the experimental animáis virtually flung themselves at the nest 
material and began to build with abnormal intensity, that is, they 
behaved in exactly the same way a normal rat would behave after having 
been deprived of nest materials for a comparable length of time; there 
was clearly a "damming up" of motor patterns which, until then, had 
been denied expression. The single behavior patterns—the searching, 
the picking up, the carrying back and the laying down of the nest mate¬ 
rial, as well as the scraping movements by which a circular nest wall is 
built, and the motor patterns of smoothing down the inside surface— 
were not different from those of the normal animáis, not even when the 
behaviors recorded on slow-motion films were compared. 

The sequence in which the cited motor patterns were carried out was 
the only way in which the behavior of the inexperienced rats actually 
differed from that of normal Controls. A normal rat never makes the 
scraping movements for piling up the nest wall before enough material 
for this is at hand, and even less does a normal rat execute the "uphol- 
sterer movements" with its forefeet, by which the inside surface of the 

9. The Deprivation Experiment 


nest is patted smooth, before the nest has an inside surface. The experi¬ 
mental animáis often exhibited the movements for piling up the nest 
wall and for upholstering after only one or two shreds of paper had been 
brought in and, left lying fíat on the floor, were not even touched as 
these motor patterns were performed. One got the impression that the 
high intensity of the nest-building activity was partially at fault for the 
topsy-turvy and disordered way in which the movements tumbled out; 
but gradually, after they had been provided with nest material for a 
longer period, the obligatory sequence of the single movements righted 
itself among the experimental animáis until each became ordered in the 
completely normal way. 

Through this non-disruptive experiment undertaken by Eibl-Eibes- 
feldt, an experiment demanding extensive knowledge of animáis, we 
learn a great deal that is by no means confined to questions concerning 
the innate, but which is much more relevant to an understanding of the 
processes of learning itself. That after a short time the rats stopped per- 
forming in the air those movements meant for piling up the nest wall 
and for upholstering can be due only to a conditioning process during 
which the reajference, that is, the feedback from stimuli which are self- 
produced through execution of the motor pattern, has a positive condi¬ 
tioning or an avoidance conditioning effect. Apparently it is "rewarding" 
for a rat, when piling up the nest wall and when upholstering the inside 
of the nest, to receive the teleonomically adequate stimulus combination 
and, very probably, it is not at all satisfying if, when executing these 
movements, the phylogenetically programmed "expectation" of extero- 
and proprioceptor stimulus configurations remains unfulfilled. 

No generalizing prediction is possible concerning where the learning 
processes of such Systems of behavior patterns are located, ñor in what 
way, organizing and adjusting, they are programmed to interact with the 
material of innate movement and reaction patterns. In the bone-burying 
behavior of a young dog, for example, the sequence of actions is appar¬ 
ently phylogenetically prescribed; what effects positive conditioning are, 
apparently, the reafferences received through digging in earth and, per- 
haps also, in a later phase of the learning process, the recovery of the 
buried booty. 

We can formúlate five rules that, in the conduct of experiments involv- 
ing the withholding of experience, must be strictly followed. 

1. What the experiment can tell us is only that a particular element of 
behavior does not need to be learned. 

2. The experimenter must possess extensive and exact knowledge of the 
action system of the animal species being tested in order to be able to 
recognize incomplete fragments of behavior patterns which, through 
the exclusión of learning processes, have been deprived of their nor¬ 
mal interconnections with other elements of behavior. Only by means 


II. The Methodology of Biology and Particularly of Ethology 

of such knowledge can the role of the learning process be ascertained 
and its ontogeny studied. 

3. The experimenter must know precisely that stimulus situation which, 
in a normal animal, releases the behavior pattern being studied, oth- 
erwise he may mistake behavior deficiencies that are caused by a 
momentary absence of stimuli for the consequences of the preceding 
withholding of experience (Riess 1954). 

4. The experimenter must have an extensive amount of experience with 
the bearing, carriage, posture, and comportment of healthy specimens 
of the species investigated, and must have a good "clinical eye" for 
the pathological consequences of an inadequate condition or a faulty 
constitution, especially for the pathological loss of intensity in partic¬ 
ular motor patterns. 

5. When experimenting with an animal reared under conditions involv- 
ing the withholding of experience, one must always begin first with 
the simplest stimulus situation possible because the learning processes 
of an animal reared under conditions of withheld experience often 
take place with such lightning-like speed at the first presentation of 
any object that afterwards one does not know which of the proffered 
attributes was responsible for the innate response. 

10. The Relatively Entirety-Independent Component 

Fortunately for our analytical endeavors, the Koehlerian definition of the 
entirety as a system of universal interrelationships does not correspond 
completely to some systems and some system parts. There are, so to 
speak, pieces in the fabric of every organism that are comparatively rigid 
and unalterable inclusions in the causal network of the rest of the system. 
Because of this, and as an example, the finished feather of a bird is no 
longer influenced by the causal network of the entirety; it is a "dead" 
element in the atomistic sense. In a restricted sense the same is also true 
for many skeletal elements, at least in their completed condition. Neither 
in form ñor in function are these parts of the whole substantially influ¬ 
enced by the entirety but, on their part, they influence the form and 
function of the entire system in a vital and decisive way. They thus stand 
in a unidirectional causal relationship to the entirety. We cali them the 
relatively entirety-independent components. 

An incorporaron of the modifying word "relative" in the definition 
of such components is necessary because such components exist in 
almost every imaginable transitional form. There are those which are 
absolutely independent and stand in a purely one-sided causal relation¬ 
ship to the system, and there are those which, during their ontogeny, are 
plástic and pliable and subject to the influence of the entirety—as are the 
bones of a vertébrate. Examples of the borderline case, of a truly and 

10. The Relatively Entirety-Independent Component 


absolutely entirety-independent component, are difficult to find; even 
the famous half ascidian that, as is known, develops from one half of the 
ovum split during its two-cell stage, is still covered with epidermis on 
that side from which the other half is missing, something that would not 
be so normally. 

The discovery of a relatively entirety-independent component pre- 
sents, within the immeasurably complex framework of the organic Sys¬ 
tem, a most welcome point of departure for causal analysis: such a com¬ 
ponent can, without committing too gross a methodological error, be 
isolated. In addition, for relatively independent components one can also 
predict that in the course of the analysis, if not exclusively then at least 
more often than not, they will appear as cause rather than as effect. For 
this reason, in research and in theory the rigid structure always repre- 
sents the Archimedean point from which the investigation proceeds. 

Every text on anatomy begins with a description of the skeleton and, 
in like manner, for research done on behavior, it was from a base of cer- 
tain not modifiable capacities of the central nervous system that analyses 
and comparative studies of behavior were legitimately launched. The 
discovery of the reflex process crystallized into the established focal 
point for the development of the physiology of the central nervous Sys¬ 
tem; the discovery of the conditioned reaction opened the way for the 
founding of the Pavlovian school of reflexology and for the development 
of behaviorism. Both of these, of course, succumbed to the error of 
explanatory monism. Finally, as was cited in the introduction and will, 
in a later section, be discussed with more thoroughness, the discovery of 
the fixed action pattern or instinctive movement by C. O. Whitman 
(1898) and O. Fieinroth (1910) provided the Archimedian point on which 
the comparative study of behavior has been built. 

The analytical possibilities which are thrown open through the dis¬ 
covery of a relatively entirety-independent component make it all too 
easy, sometimes, to forget that the methodology used when concentrat- 
ing on an isolate is legitimate only for that component under consider¬ 
aron; the researcher must remain ready to return to the otherwise oblig- 
atory methodology of analysis along a broad front, that is, to 
considerations of the entire network of amboceptory causal relation- 
ships, once he has gone beyond the boundaries of the fixed element. 

Chapter III 

The Fallacies of Non-System- 
Oriented Methods 

1. Atomism 

The sequence of research steps that has just been discussed is prescribed 
for us by the complex interacting structural framework of all living Sys¬ 
tems. The more complex a living system is, that much more strictly must 
the sequence of research steps be followed. But when dealing with sim- 
pler systems, and particularly with inanimate objects, the sequence need 
not be so strict. Let us assume that the Martian whom we have already 
used as an example has the task of analyzing a dock with a pendulum. 
It is possible to imagine that after he has understood the teleonomy of 
the object and has gone on to observe more about the speed relationships 
of one to twelve between the clock's two hands, as well as the oscillations 
of the pendulum, he would be capable, independently, of reinventing 
the mechanism of the pendulum dock. Provided he knows the laws of 
the lever and of the pendulum, not all that much inventiveness is 
required; and if he has a fair amount of luck, our Martian will, in fact, 
hit upon the same mechanism that the earthly clock-maker brought into 

That procedure followed for reconstructing a complex system on the 
basis of its observed performance and from known general lawfulnesses 
without examining its structure is described as atomistic. 

As our example indicates, atomism is by no means doomed to failure 
in principie. It merely represents a research strategy which promises that 
much less success the more complicated is the system to be analyzed. An 
organic system together with its capacities is understandably much more 
difficult to invent than is a pendulum dock. Nevertheless, there exist 
several organic functions that have been invented by technicians prior 

2. Explanatory Monism 


to their having been comprehended by biologists. Thus, for example, it 
is somewhat humiliating for us, as biologists, that the immense impor- 
tance of the self-regulating circulatory processes of homeostasis was first 
fully appreciated after such mechanisms had been devised by control 
technicians. Even in such cases where the complexity of the system 
seems to indícate that an independent construction of all of its structural 
and internal relationships is hopeless, the atomistic approach can achieve 
something of valué through providing a conceptual model that allows 
us, at least to a certain extent, to estímate the number of single functions 
to postúlate. 

2. Explanatory Monism 

While atomism as such does by no means always represent a research 
strategy which is false in principie, explanatory monism is always and 
under all circumstances a fallacy. Although explanatory monism and 
atomism are often found in combination, especially among researchers 
who think "technomorphically," they must still be thoroughly diífer- 
entiated. While the atomistic researcher is not at all blind to the fact that 
in the system he is examining many very diíferent subsystems can be 
involved, each of which responds to its own lawfulnesses, the explana¬ 
tory monist arbitrarily clutches a single part of the entirety of a living 
system and, using a simple research methodology that is readily at hand 
for determining its lawfulnesses, tries to understand the overall systemic 
function through the lawfulnesses of this single, arbitrarily selected part, 
without bothering about the other structures which form equally impor- 
tant parts of the systemic whole. In order to illustrate the error of this 
way of doing research, I return again to the Martian and to the automo- 
bile. This creature would become guilty of explanatory monism if, after 
isolating a particular mechanical part—a bolt and the nut that belongs to 
it, for example—he then turned his back on the remainder of the system 
and attempted to resynthesize the entirety from these two easily isolated 
parts. There are highly reputed schools of psychology whose members 
obstinately persist in this methodological error by claiming that that 
mechanism which enables the higher animáis to learn through experi- 
ence is the only explanatory principie on the basis of which all animal 
and human behavior can be understood. 

The noteworthy partial successes which are attained through this lim- 
ited (and I use that word in its most literal sense) approach can be 
explained by the fact that just this kind of "learning apparatus" has 
evolved in analogous ways in quite varied organisms and is, at the same 
time, a good example of that kind of partial system which can be 
described and defined as a relatively entirety-independent component. 


III. The Fallacies of Non-System-Oriented Methods 

3. Operationalism and Explanatory Monism of the 
Behaviorist School 

Nowadays a majority of civilized people do most of their day's work with 
objects that can be and have been made by man. Most people have vir- 
tually nothing whatever to do with living organisms and have forgotten 
how to deal with them. Worst of all, they have lost the respect that is due 
to all that humans are unable to make themselves. On the other hand, 
they have an exaggerated esteem that borders on veneration for their 
own technical producís and for the Sciences of physics and chemistry 
which contribute to their manufacture. This attitude with regard to the 
so-called exact Sciences is indubitably inspired by the impressive gain of 
power which humanity has acquired through analytical Science—not- 
withstanding the fact that this power has persistently proved to be the 
very opposite of beneficial. 

Because these exact Sciences (often grouped together and called "big 
Science") are based on analytical mathematics, many people assess the 
"exactitude" and, with it, the valué of every scientific result by that pro- 
portion which mathematical operations have contributed toward achiev- 
ing it. In consequence, a surprising number of people, even including 
scientists, regard counting and measuring as the only legitímate sources 
of knowledge. Therefore they try to understand mathematically the 
whole universe and everything that is in it. In other words, they 
approach the universe as if humans did not possess any cognitive capac- 
ities beyond those of counting, measuring, and calculating. 

It should be emphasized that the physicist himself is very far from thus 
limiting his own cognitive functions. No less a scientist than Werner 
Heisenberg (1969) has pointed out that the laws of logic and mathematics 
are not inherent to the extra-subjective universe surrounding us but, 
quite the contrary, are inherent in one particular cognitive function of 
man which, although it is by no means the only one, is a very great help 
to our understanding of nature. The universe, he has said, cannot cal¬ 
cúlate, but it allows itself to be calculated. Heisenberg has certainly not 
underrated the cognitive function of gestalt perception. He has called it 
"intuition," but he certainly recognizes its indispensability. 

In atomic physics and in the study of those minute particles which, at 
the present State of our knowledge, defy división, the physicist enters 
regions where most cognitive functions fail, regions where even the cat- 
egories and the forms of ideation that Immanuel Kant regarded as aprior- 
istic and indispensable prove unable to help further scientific progress. 
The evident realities of space and time come to nothing and the physicist 
has to deal with phenomena that can neither be described ñor visualized. 
They can, therefore, be defined only by means of the operations which 
produce them. Because these operations are the only way by which the 

3. Operationalism and Explanatory Monism of the Behaviorist School 


scientist can acquire knowledge concerning these realities, and because 
they can be defined only by these operations, P. W. Bridgeman (1958) 
has called their conceptualization operational. Practically all of the con- 
cepts and the terminology used in atomic physics are of this kind. 

The reason physicists use operational methods and conceptualizations 
is not because they regard them as particularly "exact" or "scientific"; 
they use them simply because they have no other methods and concep¬ 
tualizations at their disposal for work in the field under discussion. If 
and when, in everyday life, the physicist has to deal with objects that can 
be approached by the usual mechanisms of cognition, he uses these and 
would not dream of resorting to operational methods. If we should ask 
a physicist to repair some electronic apparatus built by someone else, he 
would certainly take the machine apart and determine which elements 
had been used in putting it together and how they were wired. Like any 
normal human being, he would begin his research by investigating 
structure. He would not dream of inventing a special operation, ñor 
would he stick to it as if it were the only salvation. 

As has been explained, the more complex the organic system under 
investigaron, the more indispensable is a description of structures, and, 
with this, the more multifarious are its subsystems' structures and their 
interactions. The most complex systems known to Science are the central 
nervous systems of higher animáis and man, and, in particular, the inter¬ 
actions between many such individual systems: in other words, the 
supra-individual system of a society. Paradoxically, in behavioral Science, 
in psychology and, most amazingly, quite particularly in sociology there 
prevails a fashionable tendency to ape atomic physics. The word aping 
is used intentionally because it means imitating the behavior of others 
without understanding their motivations. The physicist using the 
method of the generalizing reduction discussed in One/I/1 investiga tes 
structures not for their own sake, but in order to abstract the laws pre- 
vailing in them and to reduce these laws to more general ones. Opera¬ 
tional methods are resorted to only in those cases where there are no 
structures—when all other cognitive functions have failed. The physicist 
neither believes that structures are negligible ñor that their investigation 
is dispensable. Least of all does the physicist despise the multiplicity of 
various cognitive functions. 

All of these errors are committed by a great number of psychologists 
and sociologists. They obviously cherish the hope of finding a shortcut 
to an understanding of the most complicated organic systems by opera¬ 
tional and statistical methods. In other words, they hope to circumvent 
the weary and demanding task of gaining any causal understanding of 
the physiological machinery whose function is animal and human 

At present the most consistent representativo of this school is B. F. 
Skinner (1938, 1971), whose "empty organism doctrine" has exerted a 


III. The Fallacies of Non-System-Oriented Methods 

tremendous influence on American as well as on European psycholo- 
gists. His operational method is confined to a study, by statistical means, 
of the contingencies of reinforcement, that is, of the changes wrought in 
animal and human behavior by means of reward and punishment, but 
chiefly by reward alone. No thought is given to what it is that is being 
changed during the course of this process. A great number of present- 
day scientists believe that this method of approach is not only scientifi- 
cally legitimate but actually the solé legitimate method for approaching 
the problems of behavior. 

If this belief has been partially sustained, it is because all of those crea- 
tures so studied have, independently of each other, evolved physiologi- 
cal mechanisms that react in the same way to the experimental operations 
chosen by behaviorists. As will be explained in Part Three of this book, 
all of the phyla of animáis which have evolved a centralized nervous 
system have hit on the "invention" of feeding back to the mechanism 
initiating a behavior the consequences of its performance. In the case of 
biological success, this feedback results in reinforcement; in the case of 
failure, it has an extinguishing effect. The enormous teleonomic advan- 
tages of this kind of apparatus explain the fact that cephalopods (squids 
and octopuses) and arthropods (insects, crustácea, and spiders) all possess 
virtually identical capabilities for learning by success and failure. 

The great scientific results attained by behaviorists are not due to the 
correctness of the empty-organism doctrine, but, paradoxically, to the 
very thing whose existence this doctrine denies, that is, to the highly 
specific analogous structures convergently evolved in practically all 
organisms possessing a centralized nervous system. Without the least 
knowledge of convergent adaptation, the behaviorist school had the very 
good luck to strike, in very different organisms, strictly analogous mech¬ 
anisms and, in addition, to hit upon one which represents a "relatively 
system-independent element" in the sense explained in One/II/10. The 
process of learning by reinforcement can, therefore, be experimentally 
isolated without creating a dangerous source of error. 

Learning by reinforcement plays a highly important role in the life of 
higher animáis and humans, and for this reason the behaviorist school 
has achieved really great breakthroughs. It is necessary to emphasize this 
here because ethologists are unjustly accused of denying any merit that 
might be attributed to behaviorist research. What we reproach behavior¬ 
ists for is certainly not what they do; they do what they do in the most 
excellent manner. Our criticism refers only to their belief that there is 
nothing else in behavior to investigate. Most behaviorists shy away from 
investigating anything that is not directly connected with learning by 
reinforcement. Their program exeludes even the investigation of the 
many other kinds of learning processes. In Part Three of this book these 
very different mechanisms will be discussed. 

3. Operationalism and Explanatory Monism of the Behaviorist School 


What behaviorists exelude from the narrow circle of their interest is 
not only other learning processes, but simply everything that is not con- 
tained in the process of learning by reinforcement—and this neglected 
remainder is neither more ñor less than the whole of the remaining 
organism! The physiological mechanism achieving learning by success is 
so similar in all of the various animáis mentioned above, that not only 
the same method, but very often even the same apparatus is applicable. 
What remains uninvestigated is all that makes an octopus an octopus, a 
pigeon a pigeon, a rat a rat, or a man a man, and, most important of all, 
what makes a healthy man a healthy man, and an unhealthy man a 

When I. P. Pavlov was performing his elassie experiments on the con- 
ditioned salivating reflex of dogs, he hog-tied his experimental animáis 
in such a way that they had, within their behavioral repertory, no other 
choice than to salivate or not to salivate. This was completely legitimate 
as long as the experimenter remained conscious of the fact that he was 
investigating an artificially isolated part of a system, exactly as a phys- 
iologist does who performs experiments of excitation conduction within 
a bundle of neurites cut out of the sciatic nerve of a frog. When, on the 
other hand, behaviorists put experimental pigeons into an opaque box 
preventing perception of any information except that of when and how 
often the pigeon presses on a key, I cannot help feeling that they do not 
zvant to see the many other things undertaken by the animal, because 
they are afraid that what they see might undermine their belief in their 
own explanatory monism. The ideological reasons for this are, however, 
no subject for a textbook on ethology. 

Chapter IV 

The Comparative Method 

The first four sections of this chapter are addressed to psychologists and 
other non-biologists who may not be familiar with the facts of evolution. 
Biologists, particularly zoologists and paleontologists, should not only 
skip the first four sections, but are kindly asked to do so. This presenta¬ 
ron of the sources of our knowledge concerning evolution is extremely 
simplified and to an extent that, to the initiated, may seem impermissi- 
ble. What is important for ethologists is contained in sections 5 through 

1. Reconstruction of Genealogies 

When we speak of "comparative" anatomy or "comparative" ethology, 
the adjective has a very special connotation. It does not mean simply a 
comparison of the similarities and differences that exist, among different 
species of animáis, between the bodily forms or the behaviors—as was 
misleadingly assumed by the scientists who appropriated the word for 
use in the title of the Journal of Comparative Psychology. Comparative Sci¬ 
ence is the attempt to reconstruct, from the distribution of similarities 
and dissimilarities among living creatures, the paths along which their 
evolution has proceeded. It is necessary to discuss this method here in 
some detail because we are indebted to it for the basic discovery that gave 
rise to the Science of ethology. Ethological research started from the fact 
that there are certain sequences of movements which are as reliable as 
characteristics of species, genera, and higher taxonomic units as are any 
of the morphological characteristics used in comparative anatomy. The 
concept of homology is equally applicable to them. 

1. Reconstruction of Genealogies 


This fact alone settles a dispute which, for purely ideological reasons, 
is still going on: the age-old nature-nurture controversy. The same fac¬ 
tual basis which makes evolution a certainty proves that the form of 
homologizeable motor patterns is programmed in the genes exactly as 
morphological characteristics are. It is necessary to State and réstate this 
banal fact because, amazingly, some authors still maintain that the two 
concepts of the phylogenetically programmed and of the acquired por- 
tions of animal and human behavior are not only dispensable, but that 
they are actually false. These two concepts are no more false, ñor are they 
any more dispensable to ethological research than are the concepts of the 
genotype and the phenotype to genetics, population dynamics, or 

Whence do we obtain the knowledge that all living creatures were not, 
as the noun etymologically implies, created in their present forms but 
have obtained these through the long-lasting process which we cali evo¬ 
lution? If one puts this question to erudite non-biologists, their answers 
are very frequently that we owe our knowledge of evolution to the study 
of fossils which, enclosed in successive layers of the earth's crust, furnish 
documentary evidence of the different stages through which the devel- 
opment of life has passed on our planet. They do indeed furnish proof, 
but they are not the unique, ñor even the most important source of our 
knowledge. Even without them the main facts of life's history upon earth 
could be proved without reasonable doubt. 

The Germán mind is very orderly and it is probably due to the influ- 
ence of Germán idealism that, toward the end of the seventeenth and at 
the beginning of the eighteenth century, very sophisticated attempts 
were made to bring some sort of order into the apparently chaotic mul- 
tiplicity of life forms. One example to such attempts is the "natural Sys¬ 
tem" invented by Johann Jakob Kaup (1854) who tried to represent the 
relationships of living bird species in a diagram consisting of pentagons 
and pentagrams. He explained the choice of this particular figure by say- 
ing that the "sacredness" of the number five was derived from the num- 
ber of our senses. As shown in Figure 5, some of the angles of the pen¬ 
tagrams had to remain empty; Kaup explained this by assuming that the 
groups of birds necessary to fill these gaps were living, still undiscov- 
ered, in some remóte countries. That was in 1854. 

It is remarkable that abstruse theories such as this one could be seri- 
ously proposed more than three-quarters of a century after the Germán 
zoologist, Simón Peter Pallas, had written: 

Among all other diagrammatical representations of the system of living bod- 
ies, it would probably be best to visualize the system as a tree which, right 
at its root, divides to form a trunk consisting of the most primitive animáis 
and plants, thus forming an animal and a vegetable branch, although both 
occasionally tend to come very cióse to one another. The first branch, orig- 


IV. The Comparative Method 

Figure 5. Johann Jakob Kaup's pentagram, 1854. (Stresemann, E.: Die Entwicklung 
der Ornithologie.) 

inating with the shell-less animáis and rising up to the fishes, after having 
given off a great side-branch leading to the insects, would then lead on to 
the amphibia. And in the same manner, this branch would have to carry the 
four-legged animáis at the height of the tree's top, and it would have given 
forth, just below that, an almost equally large branch for the brids. 

This quotation, translated here from the original Germán, was taken 
from Erwin Stresemann's important book, Die Entwicklung der Ornithologie 
(1951). Stresemann comments on this passage: "With these few sen- 
tences, young Pallas has expressed a dawning idea of the real relation- 
ship between living organisms. He was the first to choose the image of 
a tree to express this relationship." 

2. Criteria of Taxa 

The image of the tree which Pallas saw with the visión of genius can be 
constructed in a manner that is as simple as it is free from hypothesis. I 
choose for its construction the taxon of the chordate animáis (Chordata) 
because the vertebrates which form its most important branch are well 
known to most readers. 

We represent the innumerable kinds of animal life forms by vertical 
lines and connect them by horizontal stripes which represent character- 
istics common to the animáis thus enclosed (Figure 6). I restrict myself to 
a small number of characteristics and try to choose those which are either 

2. Criteria of Taxa 






C! caceta rfV V , 


Aves \\¿Z 7 a 

Reptília ■ A-—* 


x jí 


Anura > h *]{ 

^ { Placenta 


Actínopte rygíi 






T un ¡cata 


V f H !f /1 1 1 i n i i 

Figure 6. Genealogical tree. An explanation is contained ín the text. 

f New temporo-maxíllary joint, 
1 maMeus and incus in ¡nner ear 

{ Mammal-líke skull structure 
and denture 

Coro if ¡catión of epidermis,_ 

Ihorny daws, 5 toes on sil legs 

/ Tetrapod extremities consist¡ng_ 

I i i^| j ¡¡i l of humerus, radius, etc, 


f Swim bladder í=Lung} r 
{ bSserial extremittes 

| Jaws, 12 cerebral nerves, -_ 

1 1 4 extremíties 

í Head with 2 eyes, labyrinth 
\ with at least 2 semicircular canals 
jSegmented musculatura, 

l notochord w¡th neural tu be-__ 

# Notochord fin embryos} 

Foregut with gífl 
slits, endostyl 

easy to describe or are so well known that description is unnecessary. 
The construction I am going to undertake becomes most convincing 
when it is executed as a three-dimensional model for which stiff wires 
are used for the animal life forms and tape is used for the characteristics 
tying some of them together. To each of the tapes a label is attached 
indicating the morphological characteristic it represents. These bindings 
are applied to arrange the great number of forms of chordates in a certain 
order; this is determined by the number of animal forms each character¬ 
istic embraces. The only way in which I might be accused of smuggling 
a hypothesis into my construction (after having promised that it would 
contain none) is this: I am arranging the characteristics most generally 
shared at the lowest level. Otherwise the construction of the tree, which 
so marvelously fits Pallas's visión, would be growing upside down, as do 
some coráis from the roof of a marine cave. However, it still would be a 

So we begin to tie together the whole bundle of Chordata forms, with 
the characteristic that is common to all of them, the chorda dorsalis or 
notochord, which is the primitive precursor of the backbone. Even at this 
early stage we encounter a difficulty: there are some animáis which pos- 


IV. The Comparative Method 

sess a notochord only in their larval State and lose it before they become 
adult; the seasquirts (Ascidia) and their relatives, in their adult forms, are 
sessile, that is, fixed to their substratum, and no one looking at them from 
the outside would suspect that they are what William Beebe (1938) jocu- 
larly called "unfortunates which have just missed becoming vertebrates." 

Considering this and looking about for similar forms of life, we find 
a number of obviously related animáis that do not and never did possess 
a notochord, but which do have a number of other structures that are 
characteristic of all true chordata. They are justly called the Protochor- 
data, "those which were there before the chordates were." They have a 
number of characteristics in common, not only with the lower chordates, 
and, because of this, significant with regard to their relationship with 
these, but which reappear, as indicators of the past, in the embryogenesis 
of all vertebrates including man. These characteristics are: a foregut 
which is perforated by gilí slits and acts as a filter, and on whose ventral 
side there is a furrow lined with ciliary epithelium collecting edible par- 
ticles and passing them on to the esophagus. This organ, the endostyl, 
still functions in the same way in larval cyclostomes and develops in 
higher vertebrates into the thyroid gland. In view of this, we attach 
below the tie indicating the notochord another and still more compre- 
hensive band representing this—and a number of other very oíd 

Another characteristic with which all chordates, with the exception of 
tunicates, are endowed, is a body with segmented musculature and a 
neural tube stretching along the body, just as the notochord does. 

Searching for further oíd and widespread properties, it strikes us that 
vertebrates have a head with two eyes and other sensory organs, such as 
an inner ear with semicircular canals serving equilibration. And to the 
surprise of some non-zoologists, we must here point to a curious animal 
which possesses a front and a rear end, as well as many fish-like char¬ 
acteristics, but has no head. This creature, "fishy" in every sense of the 
word, is Amphioxus, belonging to the sub-phylum Acrania—the headless. 
Their notochord runs forward right into the tip, and their neural tube 
ends at the front end without any sort of brain-like structure. So we must 
allow the acrania to diverge from the remaining trunk of the vertébrate 
phylum before binding our big bundle of animáis together with a new 
band that indicates a head with two eyes and a labyrinth with (at least 
two) semicircular canals—and other important organs such as a brain. 

It seems self-evident to most people that an animal with a head, a 
mouth, and two eyes should also possess an upper and a lower jaw, two 
pairs of limbs, and other well-known organs, such as two nostrils. Before 
inserting the symbolic connecting band representing these characteris¬ 
tics, we must consider a small but important group which does not pos¬ 
sess them: the "round-mouths," or Cyclostomata. They do indeed have 
a head with two perfectly vertebrate-like eyes, but they completely lack 
limbs. They have fish-like gilí slits and a large circular mouth with teeth 

2. Criteria of Taxa 


all around it, but no jaws. Unlike all other vertebrates, they have only 
two semicircular canals in their labyrinth and only ten cranial nerves 
instead of twelve. A very significant ancestral characteristic merits men- 
tioning: the foregut in the larva of some cyclostomes is constructed 
exactly as that of many protochordates and tunicates, and performs the 
same function of filtering the nutrients that are caught and conveyed to 
the esophagus by way of the endostyle which, in the later ontogeny of 
this species, is metamorphosed into a gland with internal secretion, the 
thyroid gland, which we still possess. 

All the remaining vertebrates, the gnathostomes or jawed-mouths, are 
united into a well-defined group by having upper and lower jaws, two 
pairs of limbs, twelve cranial nerves, and other characteristics in com- 
mon. This group ineludes all those vertebrates that are familiar to 

A superficial glance at this group would seem to reveal a división into 
two main branches, the fishes and the four-legged animáis—the latter, 
of course, including birds. However, our method of bracketing life forms 
according to common characteristics shows a dichotomy in quite another 
place. Older taxonomists assumed that a differentiation of true osseus tis- 
sue was the more modern acquisition, while a cartilaginous skeleton rep- 
resented the primary State. For this reason we were taught at school that 
the sharks (Elasmobranchii) were the ancestors of all other vertebrates. 
This was an error. According to more recent discoveries (Jarvik 1968), for 
example, the very oldest gnathostomes known to Science possessed true 
bones and very probably are the ancestors of higher vertebrates, while 
the sharks and rays represent an independent side branch. 

The remaining taxon, that of the bony fish (Osteichthyes), is charac- 
terized by one important organ: the swim bladder which appears either 
in the guise of a hydrostatic or in that of a respiratory organ. Very prob¬ 
ably the respiratory function was the original one and this would argüe 
for the taxon having orginated in fresh water that was not too rich in 
oxygen. Fin rays are also common to all these fishes, although the base 
of the fin is shaped in various ways; the biserial form, as shown in Figure 
7, was probably the precursor of the fin of most of the fish living now. 
In order to keep the diagram of Figure 6 as simple as possible, we can tie 
all Osteichthyes together as one branch containing three classes: first, the 
crossopterygians—today represented only by the famous coelacanth 
latimeria; second, the lungfishes (Dipneusti); and third, the teleosts, 
which designates all the other fishes. 

Among the creatures which, externally, must have looked very much 
like the crossopterygians or, for that matter, like the Australian lungfish 
(Neoceratodus), a very particular characteristic is to be found: the inner 
nostril, or choana. This characteristic ties together these oíd fishes (Rhip- 
idistians) with all recent four-footed land animáis. Some members of this 
group, besides having choanae (nostrils opening into the mouth cavity), 
also possess an arrangement of skeletal elements in their biserial limbs 


Figure 7. Above: Pectoral fins of various fishes. (a) Cladoselachii, primitive shark 
with parallel arrangement of radiáis; ( b ) Isurus, highly developed shark with 
small fin base; the basalia (dotted) are merged into three large cartilage elements: 
Pro-, Meso-, and Metapterygium; (c) Xenacanthus, paleozoic shark, with basalia 
forming an axis of the fin; ( d ) Cornubiscus (Actinopterygii, Palaeoniscoidea); ( e) 
Amia (holostei), the radiáis are arranged parallel; (/) Serranus (teleostei) with 
reduced fin skeleton; (g) Neoceratodus (lungfishes), biserial archipterygium; ( h) 
Eusthenopteron, fossil crossopterygii, arrangement of the fin skeletal part already 
resembles that of the tetrápoda. Below: Tetrapod extremity showing the carpus 
(wrist) or the tarsus (ankle). Labels at the left pertain to fore extremities, those at 
the right to hind extremities. Carpus: R Radiale, U Ulnare, the proximal wrist 
bones; T Tibiale, F Fibulare, the proximal ankle bones; 1 Intermedium; C 2 -C 4 Cen¬ 
traba; 1-5 Carpalia (hand) or Tarsalia (foot). (Remane, Storch, Welsch: Kurzes 
Lehrbuch der Zoologie.) 

2. Criteria of Taxa 


that is undoubtedly a foreshadowing of the organization of tetrapod 
limbs. In our diagram, concerned exclusively with phyletic descent, it is 
unavoidable that seemingly non-homogeneous taxa, such as the choan- 
atae which comprise fish-like forms as well as primates, are also pre- 
sented. The rhipidistia are not represented in the diagram; they ought to 
be situated between the ray-finned fish (Actinopterygii) and the newts 

All the remaining vertebrates are plainly characterized by the posses- 
sion of four legs, with an obviously homologous skeleton, consisting of 
humerus, radius, ulna, carpus, metacarpus, and phalanges in the forelegs 
and corresponding bones in the hind legs. Among the animáis with 
limbs of this kind we are again forced to mark a dichotomy because some 
of these creatures have ears with eardrums and other subservient audi- 
tory organs, while some do not. Among the amphibians, only the frogs 
have ears; ears are lacking in salamanders (Urodela). These animáis com¬ 
prise a rather mysterious group because one cannot be quite sure 
whether they have lost their auditory organs or whether these were 
"invented" after a división of the two amphibian branches. 

However that may be, all other tetrapods (four-footed animáis) do pos- 
sess ears with eardrums and inner ears constructed much like our own. 
In respect to inner ear construction, a frog and a man are more similar to 
one another than a frog is similar to a salamander, their other similarities 

A much stronger cornification of the epidermis characterizes all the 
rest of the tetrapods; a concomitant characteristic is the claw which is 
also formed of horny substance. The horny skin makes these animáis 
more independent of water; they are all distinctly terrestrial. As a result 
of this the eggs must also be made more independent of moisture: all the 
life forms united by the characteristic of skin cornification and by the 
possession of horny claws either lay large eggs with yolks and albumens, 
or they bear live young. In either case the embryo develops a very special 
organ serving respiration as well as nutrition, the so-called amnion, 
which grows out of the urinary bladder. For this reason the entire group 
is called the amniota and it comprises reptiles, mammals, and birds. 

The remainder of our genealogical tree is well known, but compli- 
cated, and for our purposes can be presented here only in a simplified 
and, as a consequence, somewhat inexact manner. A mysterious trans- 
formation of the maxillary joint is characteristic of some reptiles and of 
all mammals. The bones originally forming the joint of the jaw, that is, 
the articular bone and the quadrate bone, are disengaged from the man¬ 
dible and change their function into that of auditory organs: they are 
turned into the hammer (malleus) and the anvil (incus) in the inner ear, 
while an altogether new temporo-maxillary joint is developed. This new 
invention is common to some extinct reptiles and to all mammals. 

Leaving out of consideration the further branching off of reptiles and 
birds, we turn to the mammals. All of them possess the auditory ossicles 


IV. The Comparative Method 

derived from what originally were the bones of the maxillary articula- 
tion already mentioned; ali have hair with sebaceous giands from which 
mammary giands have evolved. Most of them give birth to live young, 
but some of them lay eggs. Those which lay eggs do not suckle their 
young because they have no teats, but they do nurse their babies after 
hatching them by means of incubation: the young feed from fíat skin 
areas excreting milk. These animáis, of which the Australian platypus 
(Ornithorhynchus) is the best known, have a cloaca, that is, a common 
opening through which uriñe, genital products, and faeces are dis- 
charged, exactly as reptiles and birds have. Deceptively, these "cloacata" 
also have a bilí and no teeth, which made naive taxonomists regard them 
as a "missing link" between mammals and birds. 

All other mammals bear live young, but are divided into two sharply 
distinct groups on the basis of the stage of development at which these 
young are delivered and by the way the embryos are nourished. A vast 
majority of surviving mammalian forms belong to the group of Placen- 
talia (or Eutheria). The embryo develops a special alimentary organ, the 
placenta, which permits it to absorb nutrients from the mother's body. 
This makes it possible to postpone the date of birth considerably and 
thus to produce more completely developed offspring. The compara- 
tively small group of non-placentalian mammals, the marsupials, have to 
deliver their young at a much less developed stage and at a much smaller 
size. The neonate of the largest marsupial, the great red kangaroo 
(Macropus rufus), is the size of a bean; that of the blue whale (Sibbaldus 
musculus) is six meters long. Everyone knows the size of a young calí or 
deer at birth. The neonate marsupial, which looks very much like a pla- 
centalian embryo, enters the pouch (marsupium) of its mother and finds 
the teat, whereupon the epithelium of its lips fuses with that of the teat 
so as to form a hermetic connection through which milk is pumped into 
the baby's intestinal tract. In a manner of speaking, the babies attach 
themselves to a secondary umbilical cord. 

Dispensing with a more detailed description of the rather well-known 
part of the genealogical tree comprising mammals and birds, we can pro- 
ceed to the question concerning what conclusions can be drawn from the 
fact that the immense multiplicity of living creatures, by a method 
devoid of any preconceived hypothesis, can be arranged to form a dia- 
gram which automatically and without any constraint assumes the form 
of a tree. This diagram could be extended and vastly enriched by adding 
an immense number of further characteristics that would, by fitting in 
without difficulties or contradictions, bring the probability of the conclu¬ 
sions to be drawn near certainty. Even without the additional testimony 
of fossils, it can only be regarded as an historical fact that evolution has 
taken place. Historians would consider it an affront if somebody were to 
demand of them further proof supporting their "theory" that Caesar or 
Charlemagne had existed. Evolution is better documented than that! 

4. Documentation Through Fossils 


3. The Hypothesis of Growth 

Wherever in nature we encounter the form of a tree, we automatically 
assume that it has grown. A plant, a colony of coráis, a tree, the antlers 
of a deer, all have started their existence as a single sprout which, grow- 
ing longer, has divided into a number of branches. Even the ice crystals 
ornamenting our window panes in coid weather have done so. From the 
facts which have been mentioned and from an immense number of facts 
not mentioned, we deduce a hypothesis explaining their relationship: we 
assume that the characteristics which are common to a certain number of 
living forms are those which their common ancestor possessed. We say 
that these characteristics are homologous. 

The particular form of our genealogical tree of chordates is explained 
by the assumption that all of them are descended from creatures pos- 
sessing the characteristics identified with the lowest crossband in our 
diagram. The term "descended," by the way, is quite misleading; 
"ascending" would be, on the whole, much better. Furthermore, we 
assume that if only a small number of forms possess a certain character- 
istic, then this represents a new accomplishment in evolution, a new 
"invention" achieved by the method of random genetic change and nat¬ 
ural selection. It is the coming-into-existence of something new that is 
symbolized by the growth of the tree in our diagram. This assumption is 
only reliable when the new characteristic is not brought about by mere 
simplification, let alone by the disintegration of a structure which 
already existed. In fact the "newness" must be documented by reliable 
evidence proving that this particular characteristic has originated by 
complication and differentiation and as a systemic property of the organ- 
isms that did not exist previously. This proof can indeed be proffered 
and found to be beyond any reasonable doubt whenever structures that 
are characteristic of only a special branch of the genealogical tree can be 
traced back to other structures that were simpler and were found in a 
greater number of forms at a lower level. Examples of such structures are 
the tetrapod limb evolving from the rhipidistian fin (Figure 7h) and the 
bird's feather evolving from the reptile's scale. 

Considering all that has been said in this and in the preceding section, 
the growth of the genealogical tree—evolution—would be a near cer- 
tainty on the basis of comparison alone, even without the additional tes- 
timony of fossils. 

4. Documentation Through Fossils 

If our interpretation of the distribution of older and of more recent char¬ 
acteristics is correct, the more widely distributed characteristics should 
appear first, in the lower, that is, in the older strata of our planet's crust. 


IV. The Comparative Method 

The more special characteristics should make their appearance in the 
same sequence of geoiogic leveis as they do in the levels of our tree dia- 
gram which, constructed on a basis of comparison, has automaticaiiy 
made them appear—and indeed they do! 

There is no fossil documentation recording the evolution of proto- 
chordates. Because none of these animáis possesses any skeletal elements 
hard enough for fossilization, this is to be expected. During the Silurian 
a considerable number of fossil fish appear, all of which belong to the 
cyclostomes, as very thorough investigations by Stensió (1927) have 
revealed. The forms of these fishes' bodies are not always like those of 
lampreys or other modern cyclostomes, all of which have evolved, in 
adaptation to their special ecological niches, an elongated, eel-like shape. 
The oíd cyclostomes varied in form; some were spindle-shaped and prob- 
ably led a free-swimming life; others, like the cephalaspids, were dorso- 
ventrally flattened and heavily armoured with osseous plates and cer- 
tainly living on the bottom of the sea. None possessed limbs or jaws. 
Very probably some or most of them lived by filtering nutrients from the 
water by means of the foregut, as the protochordates as well as the larvae 
of some extant cyclostomes do. 

Gnathostomes, that is, fish posssessing jaws (palatoquadrate and man¬ 
dible), arise during the Silurian, but unambiguously recognizable fossils 
are not found before the Devonian. Sharks and ray-finned fish appear 
approximately at the same level. According to Nelson, the Acanthodii 
are to be regarded as the most primitive gnathostomes, but opinions dif- 
fer as to whether they are more closely related to the sharks (Elasmo- 
branchii) or to the bony fish (Actinopterygii). The first fossil fish clearly 
belonging to this group show an unmistakable similarity to certain lung- 
fishes (Ceratodiformes) as well as to the lobe-finned fishes (Crossopter- 
ygii) to which the famous coelocanth Latimeria also belongs. As early as 
the lower Devonian, the branches of sharks and ray-finned fishes (Acti¬ 
nopterygii) are clearly divided and, almost at the same time, the choanae 
are "invented" by the Rhipidistia. Among the Rhipidistia, Osteolepis 
hints at the origin of tetrapods through the form of its teeth, which are 
similar to those of amphibia, while the closely allied Eusthenopteron pos- 
sess the limb skeleton shown in Figure 7h. Decisive steps of evolution 
often follow very quickly, one after another—which means within only 
a few million years. 

One structural property of our hypothesis-free diagram will strike the 
attentive reader: few of the divisions result in two branches of equal 
thickness, that is, of an equal number of living representativos. The 
"invention" of jaws and limbs separates the very few still-living cyclo¬ 
stomes from all other vertebrates, that of the amnion gives origin to the 
host of reptiles, birds, and mammals, leaving behind the comparatively 
tiny group of amphibia. The only dichotomy which divides two approx- 

4. Documentation Through Fossils 


imately equivalent branches is the one between the fishes and the ter- 
restrial tetrapods. The reason for this phenomenon is rather obvious: 
these two branches do not compete; the conquest of dry land by the 
tetrapods did not impair the chances for survival of the aquatic 

In a majority of cases, however, the coming-into-existence of an 
unprecedented new physical organization, like that of the palatoquad- 
rate with jaws, or of limbs, brought such overwhelming advantages to 
those which possessed them that those which did not succumbed to their 
competition. The obsolete forms occasionally became extinct; occasion- 
ally they succeeded in eking out modest existences in far-fetched ecolog- 
ical niches where they were exempt from competition with the more 
modern organisms. Lampreys and other surviving cyclostomes are an 
example of this success. They survive in the struggle for existence with 
the gnathostomes in a way analogous to that of sailing ships which still 
exist in Coastal waters and for local trade, weakly competing with diesel 
engine motorboats. 

An organ which has had an interesting evolutionary history is the 
swim bladder. The cióse similarity between the oldest actinopterygian 
(ray-finned) fishes such as Cheirolepis and contemporary crossopterygian 
(lobe-finned) fishes as well as lung fishes (Dipnoi) makes it probable that 
all these forms originated in fresh water that was not too rich in oxygen 
and which made an accessory respiratory organ desirable. Also the 
palaeoniscids, typical ray-finned fish of the periods following the Devon- 
ian, lived in fresh water. It is safe to assume that the swim bladder 
evolved first under the selection pressure of breathing and only later 
developed the function of a hydrostatic organ. 

While the sharks (Elasmobranchii) have always been marine animáis, 
the Actinopterygii conquered a place in the sea in spite of the elasmo- 
branch's competition. They succeeded in doing so because the swim 
bladder permitted them to evolve a much more solid skeleton; without 
the help of the hydrostatic organ, such a skeleton would have increased 
their specific weight by too much. Also, they can twist their bodies to 
perform the undulating swimming movements common to chordates 
from the amphioxus upwards, with a much shorter "wave-length" than 
sharks can. This implies a tremendous increase in their propelling forcé. 
It is easily possible for a strong man to grasp and hold a shark the size of 
an adult arm, but just try this with a grouper (serranid) of the same size. 
The máximum speed attainable by a fish is, of course, directly propor- 
tional to the strength of its propulsión. It is easy to understand why the 
bony fish, on their return to the sea, found it easy to compete with 

As early as during the late Devonian the vertebrates conquered "dry 
land," although they were certainly confined, at first, to rather moist 


IV. The Comparative Method 

areas. From this period we possess fossil documentation of bizarre forms 
intermedíate between crossopterygians and amphibians, such as Ich- 
thyostega which possessed typical tetrapod limbs but, at the same time, 
had a tail with a fish's caudal fin. The rhipidistian Osteolepis had true 
nostrils and choanae, as well as folded enamel layers on its teeth and a 
number of structural characters of the labyrinthodonts, amphibians 
which also make their first appearance during the Permian period. 

During the Carboniferous, the epoch preceding the Permian, the divi¬ 
sión between amphibians and reptiles was accomplished. This división 
is not at first as sharply defined as it is between the living representatives 
of these two classes because the characteristics which carne to distinguish 
them are not discernible in fossils and because they are not unequivo- 
cally distributed among the two classes. In the upper Carboniferous for- 
mations and in the lower Permian, a group of stegocephalians has been 
preserved which, in the structure of their skulls, closely resemble the 
first true reptiles—the cotylosaurians—but which, as juveniles, pos¬ 
sessed a structure of the occipital región of the head clearly indicating 
that they were breathing through external gills—as many amphibians 
do during the larval stage. Nevertheless there is little doubt that the great 
innovations which made the Tetrápoda really independent of water and 
which were accomplished at the transition from the Carboniferous to the 
Permian, the cornification of the epidermis and the large egg having a 
water-impermeable shell and containing a large yolk and an embryo 
with an amnion, all represent "innovations" made during that eventful 

Fossil documents show that the división of the reptiles into the two 
important branches, one leading to the birds and the other to the mam- 
mals, took place at a very early time, at the level of the cotylosaurians, 
which represent the most primitive of all true reptiles. The further evo- 
lution of the branch leading to reptiles and birds is complicated and con- 
cerns many extinct forms not familiar to non-zoologists, and will not be 
discussed further here. The group of pelycosaurians, originating in the 
upper Carboniferous and ending in the upper Triassic, evolved a grad- 
ually increasing number of characteristics foreshadowing mammals. In 
the Permian they gave rise to the therapsids, which resembled mammals 
very closely indeed with regard to the structure of the skull and partic- 
ularly to that of the jaw articulation. They are found, in some 300 species, 
up to the lower Triassic. 

The first remnants of true mammals are found in the upper Triassic. 
These remnants consist mainly of jaw bones and teeth which, because of 
their solidity, are the skeleton elements most likely to become fossilized. 
They clearly belong to mammals, yet they belong with equal unambi- 
guity to neither the marsupials ñor to the placentalians but very probably 
to common ancestors of both. It is certainly significant that these rem¬ 
nants of the earliest mammals are found in the same strata as those of 

5. Homology and Its Criteria 


the most mammal-like reptiles: in the upper Triassic of the Cape of 

From this time onwards the mammals continued to lead an inconspic- 
uous existence as small insectivores and externally shrew-like animáis. 
They outlived the dinosaurians whose heyday lasted throughout the 
Jurassic and the Cretaceous. At the dawn of the Tertiary period the mam¬ 
mals suddenly proliferated and became the most successful group of ter- 
restrial animáis. Very probably this development was closely connected 
with the conquest of dry land by the flowering (phanerogamous) plants. 
The phanerogam's seed, enclosing a large store of nutrients and a com- 
paratively large embryo, is curiously analogous to the amniote's egg. The 
rich source of food offered by the leaves and seeds of flowering plants 
offered innumerable opportunities for divergent adaptation. 

Even the extremely limited selection of fossil documents referred to 
here should suffice to convince anybody that the "theory" of evolution 
is no longer a theory at all but plain history, and a history that is better 
documented than is any part of our own human history, either by means 
of the written word or through the testimony of oíd cultural relies. Never 
once has a fossil been found to indicate the existence of any animal at a 
lower level in the genealogical stratification than that in which its char- 
acteristics, according to the inferences to be drawn from our genealogical 
diagram, are expected to appear. And as will be remembered, the dia- 
gram was constructed without any hypothesis and on the exclusive base 
of a systematic evaluation of characteristics. In Figure 6, the horizontal 
lines corresponding to geological periods can be drawn in without ever 
leading to a contradiction between the deductions of comparative mor- 
phology and the findings of stratigraphical paleontology. 

5. Homology and Its Criteria 

For the purpose of comparative ethology, a very simple definition of 
homology is sufficient. This definition has already been stated in the 
Introductory History on page 3: Characters of two or more species are 
homologous when they owe their similarity to the common descent from 
ancestors possessing them. Adolf Remane (1952, 1959) has added a num- 
ber of additional criteria serving to ascertain the homology of compara¬ 
ble organs in different species. As these have been conceived for the pur¬ 
pose of determining the homology of morphological structures, not all 
of them are applicable in the comparative study of behavior. 

The first and, virtually, still the most important among Remane's cri¬ 
teria of homology is that of "special quality" (spezielle Qualitat). This 
means that the homology of two structures is that much more certain the 
greater are the number of coinciding details both of them possess, the 
more complicated these details are, and the more precise their agree- 


IV. The Comparative Method 

ment. What has been said in One/II/3 about the cognitive functions of 
perception should explain that this criterion of special quality is not only 
accessible to direct observation but that, also, our gestalt perception is 
the main source of the knowledge on which assertions concerning the 
"special quality" of structures are based. Our direct perception is able to 
subsume, within one unmistakable quality, a greater number of details 
than any rational computation can. For this very simple reason any 
attempt at a "numerical taxonomy" is doomed to complete failure— 
which it well deserves because of the ridiculous epistemological error on 
which it is based. For reasons that are easily understood, the criterion of 
"special quality" plays an important role in the comparative study of 

The second of Remane's criteria of homology is that of the position of 
a structural element in relation to the adjacent ones surrounding it. A 
bone in the skull of a vertébrate, for instance, can have become very 
small by a process of reduction and still remain recognizable through its 
relation to surrounding bones. With regard to behavior patterns, 
between which only a temporal relationship is possible, the criterion of 
position must be used with some caution. As the comparative study of 
courtship movements in dabbling ducks has shown, the same motor pat¬ 
terns can appear in different sequences or couplings in different species. 
The criterion of position would here be misleading. There are other 
instances, however, in which it is significant. The taxonomic dignity of 
criteria has to be assessed in much the same way as that of single char- 
acteristics, as shall be discussed in Section III/10. 

The third of Remane's criteria of homology comprises the existence of 
transitional forms between two characteristics whose homology is to be 
ascertained. The fin skeleton of Eusthenopteron (Figure 7h) may serve as 
an example of a transitional form between the biserial fin skeleton of a 
crossopterygian (Figure 7g) and that of the tetrapod limb (Figure 9). As 
Wolfgang Wickler (1970) has emphasized, an assertion made on the basis 
of this criterion can, in some instances, be deduced from the correct use 
of the first two criteria of Remane. On the other hand, there are cases in 
which the existence of transitions can justify the assertion of homology 
even when the first two criteria are not applicable. A detailed series of 
transitions can prove the homology of characteristics which are neither 
qualitatively similar to one another ñor positioned in a similar way. The 
criterion of transition is the more valuable the more certain the cióse 
relationship of the species concerned can be deduced from still other cri¬ 
teria. The homology of certain courtship patterns of Anatini, which are 
often qualitatively changed beyond recognition by the process of ritual- 
ization and which furthermore are in no way characterized by position, 
can still be ascertained with tolerable assurance exclusively on the basis 
of transitional forms. 

6. The Number of Characteristics as a Criterion of Homology 


6. The Number of Characteristics as a Criterion of 

Remane's criterion of "special quality" can be regarded from a quantita- 
tive point of view. The number of details, and the degree to which the 
coinciding characteristics of two living species are complicated, do not, 
after all, mean anything but the greater probability that both are indeed 
closely related to one another. Thus expressed, the criterion can, in a 
manner of speaking, be inverted. It is perfectly legitimate and correct to 
say: when two or more forms of life are found to agree in a very great 
number of characteristics and to be different from each other in but very 
few, it is safe to assume that the majority of characteristics are homolo- 
gous and the few others are not. If, conversely, some creatures are similar 
to each other through only a very few characteristics, while each of them 
is tied to another group by a great number of characteristics, it is equally 
safe to assume that these few coinciding characteristics owe their exis- 
tence to what is called convergent evolution—which will be discussed 
in the next section. 

In the diagram of Figure 6, a loop is drawn around sharks, fish, 
amphibia, reptiles, birds, and mammals and is marked "four-limbed." It 
may have struck the attentive reader that some of the creatures thus 
entwined have no limbs. Moray eels among fish, gymnophionids among 
amphibia, slow-worms (Anguidae) among reptiles and, finally, snakes 
lack any kind of limb. By what right, if by any, do we assume that the 
limblessness of all these creatures is not a primary characteristic as it is 
in cyclostomes? Why do we classify each of them with another group of 
creatures that does have four limbs? 

Our assertions are based on a very simple consideration of probabili- 
ties. If the lack of limbs in the four groups mentioned above were a pri¬ 
mary feature, we should have to assume that a sepárate fine of descent 
leads from the cyclostomes to each of these legless groups of animáis. 
This is extremely unlikely because the moray is proved by innumerable 
other characteristics to be a bony fish, the slow-worm a lizard, and so on. 
The alternative explanation would be that all those many characteristics, 
which each of these legless creatures has in common with vertebrates 
other than cyclostomes, owe their agreement to convergent evolution— 
which is of an overwhelming improbability. 

The more characteristics that are known, the greater the reliability of 
this computation. With fossils, for which only skeletal characteristics are 
available, the paucity of these can lead to errors when relating particular 
fossils to certain groups. A well-known paleontologist, Schindewolf 
(1936), related the Jurassic ichthyosaurians to whales, in particular to dol- 
phins, although the former show through as many characteristics that 


IV. The Comparative Method 

they are reptiles as the cetaceans possess to prove that they are mammals. 
The external similarity between the two marine tetrapods rests on a very 
limited number of convergent characteristics, such as the streamlined 
form of the body, the fin-like limbs, the bill-like snout armed with small 
and uniform teeth, and others. 

7. Convergent Adaptation 

To be certain of homologies is all the more important because there is 
another way by which similarities can arise; these similarities could be 
mistaken for homologies and this would lead to erroneous conclusions 
concerning phyletic descent. This second way, which has already been 
hinted at, is convergent evolution: two or more groups of animáis can, 
independently of one another, hit on the same way to cope with a certain 
environmental problem. Hawks (Falconidae), swifts (Cypselidae), swal- 
lows (Hirundini) and other birds have evolved fálcate wings and a 
streamlined body in order to pass easily through the air, while sharks 
(Elasmobranchii), innumerable bony fishes of different orders (Teleos- 
tei), ichthyosaurians (Reptiles), whales (Cetácea), and dolphins (Mam- 
malia) have found the same adaptation for swift movement in the heav- 
ier médium of water. Another classic example of morphological 
adaptation to a certain type of movement is represented by some large 
moths (Saturnidae), on one hand, and hummingbirds (Trochilidae) on 
the other; both have specialized in sucking néctar from flowers while 
hovering in the air in front of them. While doing this, these two 
extremely different kinds of creature look surprisingly similar and, on 
even closer examination, so do the proportions of their bodies and wings. 

In some cases convergent adaptation can lead to results which are so 
excessively similar that, without a very cióse examination, one could be 
tempted to apply Remane's first criterion of special quality and take for 
homologies these results of convergent adaptation. Figure 8 shows lon¬ 
gitudinal sections made through the eyes of a vertébrate and a cephalo- 
pod, a group which belongs to the mollusks, that is, to the same phylum 
as snails (Gastropoda) and clams (Bivalvae). The caption suífices to indi- 
cate the extreme similarity concerning even minute details. Nonetheless 
this similarity is the result of convergency, as can be shown by means of 
other criteria—by comparing embryogenesis, to ñame but one of them. 

Wherever such a similarity of characteristics is found which definitely 
can not be explained by common ancestry, convergent adaptation to a 
common function can be assumed with an overwhelming degree of cer- 
tainty. The probability of two forms of life evolving, by sheer coinci- 
dence, a certain number of identical characteristics, can be calculated. It 

is equal to ¡v n being the number of similar or identical characteristics. 

8. Analogy as a Source of Knowledge 


Figure 8. Detailed analogy in two independently evolved light-perceiving 
organs. Left, the eye of an octopus; right, the eye of a man; co, cornea; ci, Corpus 
ciliare; m.ri, musculus ciliaria; i, iris; r, retina. (Lorenz: "Analogy as a Source of 

The probability of ten characteristics being present by puré chance in 
two unrelated forms of life is therefore 1:512. 

Characteristics and organs which, in different forms of life, owe their 
similarity to the processes of convergent evolution are called analogous. 
The study of analogies is of particular importance to the investigation of 
social behavior in animáis and man. 

Exclusión of homology is a prerequisite for ascertaining analogies. On 
the other hand, the assumption of homology is the more certain, the 
more different is the function of the organs in which it is found. If we 
find the same skeletal elements in the forelimb of tetrapods, although it 
performs a very different function for each of them, their homology 
needs no further confirmation. 

8. Analogy as a Source of Knowledge 

Ethologists are often accused of drawing false analogies between animal 
and human behavior. However, no such thing as a false analogy exists: 
an analogy can be more or less detailed and henee more or less inform- 
ative. Searching assiduously for a "false" analogy, I found a couple of 
technological examples within my own experience. I once mistook a ship 
mili for a stern-wheeler. A vessel was anchored on the banks of the Dan- 
ube near Budapest. It had a little smoking funnel and at its stern an enor- 
mous, slowly turning paddle wheel. On another occasion, I mistook a 
small electric power plant, consisting of a two-stroke engine and a 
dynamo, for a compressor. The only biological example that I could find 


IV. The Comparative Method 

concerned a luminescent organ of a pelagic gastropod, which was mis- 
taken for an eye because it had an epidermal lens and, behind this, a 
high cyiindrical epithelium connected with the brain by a nerve. Even 
in these exampies the analogy was faise only with respect to the direction 
in which energy was transmitted. 

There is, in my opinión, only one possibility for an error that might 
conceivabiy be described as the "drawing of a faise analogy" and that is 
mistaking a homology for an analogy. 

This fact becomes important in the study of behavior. Not being vital- 
ists, we hold that any regularly observable pattern of behavior which, 
with equal regularity, achieves survival valué is the function of a sensory 
and nervous mechanism evolved by the species in the Service of that par¬ 
ticular function. Of necessity, the structures underlying such a function 
must be very complicated, and the more complicated they are the less 
likely it is, as we already know, that two unrelated forms of life should, 
by sheer coincidence, have happened to evolve behavior patterns which 
resemble each other in a great many independent characteristics. 

A striking example of two complicated sets of behavior patterns evolv- 
ing independently in unrelated species, yet in such a manner as to pro¬ 
duce a great number of indisputable analogies, is furnished by the 
behavior of human beings and of geese when they fall in love and when 
they are jealous. Time and again I have been accused of uncritical anthro- 
pomorphism when describing, in some detail, this behavior of birds and 
people. Psychologists have protested that it is misleading to use terms 
such as falling in love, marrying, or being jealous when speaking of 
animáis. I shall proceed to justify the use of these purely functional con- 
cepts. In order to assess correctly the vast improbability of two compli¬ 
cated behavior patterns in two unrelated species being similar to each 
other in so many independent points, one must envisage the complica- 
tion of the underlying physiological organization. Consider the mini- 
mum degree of complication which even a man-made electronic model 
would have to possess in order to simúlate, in the simplest possible man¬ 
ner, the behavior patterns here under discussion. Imagine an apparatus, 
A, which is in communication with another apparatus, B, and which 
keeps checking continuously whether or not apparatus B communicates 
with a third apparatus, C, and which, furthermore, on finding that this 
is indeed the case, does its utmost to interrupt this communication. If one 
tries to build models simulating these activities, for example, in the man¬ 
ner in which Grey-Walter's famous electronic tortoises are built, one 
soon realizes that the minimum complication of such a system far sur- 
passes that of a mere eye. 

The conclusión to be drawn from this reasoning is as simple as it is 
important. Since we know that the behavior patterns of geese and men 
cannot possibly be homologous—the last common ancestors of birds and 
mammals were extremely primitive reptiles with minute brains and cer- 

8. Analogy as a Source of Knowledge 


tainly incapable of any complicated social behavior—and since we know 
that the improbability of coincidental similarity can only be expressed in 
astronomical numbers, we know for certain that it was more or less iden- 
tical survival valué which caused jealousy behavior to evolve in birds as 
well as in man. 

This, however, is all that the analogy is able to tell us. It does not tell 
us wherein this survival valué lies—although we can hope to ascertain 
this through observations of and experiments with geese. It does not tell 
us anything about the physiological mechanisms bringing about jealousy 
behavior in the two species; they may well be quite different in each case. 
Streamlining is achieved in the shark through the shape of the muscu- 
lature, in the dolphin by means of a thick layer of blubber, and in the 
torpedo with welded Steel plates. By the same token, jealousy may be— 
and probably is—caused by an inherited and genetically fixed program 
in geese, while it might be determined by cultural tradition in humans— 
though I do not think it is, at least not entirely. 

Limited though the knowledge derived from this kind of analogy may 
be, its importance is considerable. In the complicated interactions of 
human social behavior, there is much that does not have any survival 
valué and never did have any. So it is of some significance to know that 
a certain recognizable pattern of behavior does, or at least once did, pos- 
sess a survival valué for the species; in other words, that it is not patho- 
logical. Our chances of discovering wherein the survival valué of the 
behavior pattern lies are vastly increased by finding the pattern in an 
animal with which we can experiment. 

When we speak of falling in love, of friendship, of personal enmity or 
of jealousy in these or in other animáis, we are not guilty of anthropo- 
morphism. These terms refer to functionally determined concepts, just as 
do the terms legs, wings, eyes, and the ñames used for other bodily struc- 
tures that have evolved independently in different phyla of animáis. No 
one uses quotation marks when speaking or writing about the eyes or 
the legs of an insect or a crab, ñor do we when discussing analogous 
behavior patterns. 

However, when using these different kinds of terms, we must be very 
clear as to whether the word we use at a given moment refers to a con- 
cept based on functional analogy or to one based on homology. The word 
"leg" or "wing" may have the connotation of the first kind of concept in 
one case and of the second in another. Also, there is the third possibility 
of a word connoting the concept of physiological, causal identity. These 
three kinds of conceptualization may coincide or they may not. To make a clear 
distinction between them is particularly important when one is speaking of 

A homologous behavior pattern can retain its ancestral form and func- 
tion in two descendants, and yet become physiologically different. In 
mammals motor patterns of locomotion that are demonstrably caused 


IV. The Comparad ve Method 

primariiy by endogenous impulse production and central nervous coor- 
dination (von Holst 1969-70) can come under the control of higher loci. 
What has been, primariiy, a step taken with the foreleg may become a 
voluntary movement. Jakob von Uexküll showed that in adult scypho- 
medusae the contraction of the umbrella was dependent on a "reflex" 
being released within the marginal bodies whenever the expanding 
umbrella suddenly reached the limit of its elasticity. As he said in his 
poetic manner: "It hears nothing but the tolling of its own bell." More 
recently, the marginal bodies have been shown to possess an autono- 
mous generation of activity. Reflexes, however, do not seem to have been 
completely eliminated; I myself once removed all but two opposite mar¬ 
ginal bodies in a large Scyphostoma pulmo. Its umbrella continued to púl¬ 
sate although a certain "lagging behind" could be noticed in those parts 
of the umbrella located farthest from the remaining marginal bodies. 
When I put both hands on the umbrella just outside the marginal bodies 
and allowed it to expand very slowly, the pulsation stopped. Pulsation 
could be started again by means of a slight tap from the outside. At that 
time (more than fifty years ago) I had no idea about endogenous impulse 
production and so I did not wait for the umbrella's pulsation to begin 
again spontaneously. The specimen in question had been washed ashore 
and its lack of spontaneity may have been the result of its being at the 
point of dying. In Scyphostoma a stable position in the water is main- 
tained by the different specific weights of the umbrella and the remain- 
der of the body. The functions of the marginal bodies are not necessary 
for the maintenance of equilibrium. In Hydromedusae this is different; 
when the small medusae of Stauridium (Cladonema) swim at all, they 
start as unpredictably as a housefly taking off, and they then move 
upward with an irregular zigzag motion, obviously overcompensating, 
with each contraction, for the deviation from the vertical incurred dur- 
ing the preceding contraction. In Gonionemus and Craspidacusta the 
equilibrating movements are not noticeable. 

A homologous motor pattern may retain its original physiological 
causation as well as its external forms, yet undergo an entire change of 
function. The motor pattern of "inciting" that is common to the females 
of most Anatidae is derived from a threatening movement and its primary 
function is to cause the male to attack the adversary indicated by the 
female's threat. In some species it has lost this function entirely; in the 
goldeneyes, for instance, it has become a puré courtship movement of 
the female. 

Two non-homologous motor patterns of two related species may, by a 
change of function, be pressed into the Service of the same survival 
valué. The preflight movement of ducks, an upward thrust of head and 
neck, is derived from an intention movement of flying, while the cor- 
responding signal of geese is derived from a displacement shaking of the 
head. When we speak of "preflight movements of Anatidae" we form a 
functional concept that embraces both. 

10. Systematics and the Need for Great Numbers of Characteristics 


These examples are sufficient to demónstrate the importance of keep- 
ing functional, phylogenetic and physiological conceptualizations 
clearly sepárate. Ethologists are not guilty of "reifications" or of illegiti- 
mate anticipations of physiological explanations when they form con- 
cepts that are only functionally defined—such as, for instance, the con- 
cept of the IRM, the innate releasing mechanism. They are, in fact, 
completely aware that this function may be performed mainly by the 
sensory organ itself, as has been demonstrated in the cricket by Regen 
(1924) and in the females of some mosquitoes, which respond exclusively 
to the sound frequencies produced by the males. In the eye of the frog, 
as Lettvin and his co-workers (1959) have demonstrated, a certain prelim- 
inary filtering of stimuli is performed by the retina which, however, 
actually is a part of the brain itself. 

9. Homoiology 

Besides homology and analogy, no other explanation can be found for 
the appearance of similar-to-identical characteristics in different forms of 
life. There are, however, mixtures of the two; there exist similarities 
which are caused by both. The wings of the flying saurian and a bat 
shown in Figure 9 are indubitably homologous with regard to their bony 
elements, while the flying membrane found in both has certainly been 
evolved convergently by reptiles and by mammals. Another example is 
furnished by the flippers of ichthyosaurians and whales: both possess 
homologous bones, but in both convergent adaptation to the same func¬ 
tion has caused these bones to become shortened and flattened and, in 
the metacarpal bones as well as in those of the digits, has caused the 
middle part (diaphysis) to sepárate from the ends (epiphysis) so as to cré¬ 
ate three bones out of one (Figure 9, 3). All these analogies serve to 
broaden the flipper and to make it, as a whole, flexible. The results of 
such superpositions of convergent adaptations on pre-existing homolo- 
gies are called homoiologies. 

10. Systematics and the Need for Great Numbers of 

For obvious reasons, homoiologies are most abundant when two kinds 
of animáis, closely related to one another in the first place, become even 
more similar to each other by convergent adaptation. The dual effect of 
homology and analogy can become very confusing and has often misled 
taxonomists into an erroneous grouping of non-related forms. A striking 
example of this is the genus Aquila , the "eagles" of older ornithology. All 
raptors exceeding a certain size were, at that time, subsumed under this 
genus. Size in these birds should be considered an adaptation to the 


IV. The Comparative Method 

Figure 9. Anterior limbs of vertebrates. (1) Jurassic flying reptile; (2) bat; (3) 
whale; (4) sea lion; (5) mole; (6) dog; (7) bear; (8) elephant; and (9) man. The 
humerus and the metacarpal bones are tinged in black, the carpal bone in grey. 
(Lorenz: "Analogy as a Source of Knowledge.") 

catching of comparatively large prey, and this explains why raptors of 
different genera, even of different families, have convergently evolved 
a number of conspicuous characteristics. All of them have a large, heavy 
head with a strong bilí, large eyes shadowed by bony ridges—which 
simúlate a savage frown, short legs with thick toes and large talons, a 
short tail, and broad wings—which have to do with the necessity of swift 
flight, of stooping, and of carrying away heavy loads. Even in the oth- 
erwise excellent Tierenzyklopadie of the Urania Tierreich, published in 
1975, the "eagles" are treated as a phyletically coherent group, although 
they actually belong to four different families. The golden eagle (Aquila 
chrysaetus) and the imperial eagle (Aquila heliaca) originated from the buz- 
zards (Buteonini); the sea eagle (Haliaetus albicilla), the bald eagle (Haliaetus 
leococephalus) and the Japanese giant sea eagle (Haliaetus pelagicus) belong 
to the family of the kites (Milvini); while the booted eagle (Hieraetus pen- 
natus), Bonelli's eagle (Hieraetus faciatus), and a number of huge tropical 
forms such as the harpy (Thrasaetus harpyia) are very probably descended 

10. Systematics and the Need for Great Numbers of Characteristics 


from the goshawk family (Accipitrini). If one evaluates all of the less con- 
spicuous details, of morphology as well as of behavior, the relationship 
of the several alleged eagles to these three groups can hardly be doubted. 
All the zoo people and all the falconers with whom I have spoken, who 
knew the groups in question by cióse and unprejudiced observation, 
regarded the grouping here proposed as obvious. Taxonomists, who 
work exclusively in museums, whom oíd Alfred Edmund Brehm 
despised so much he called them "Balgforscher" (skin scientists), tend to 
over-assess the importance of skeleton structures and of measurements. 
As my teacher, Erwin Stresemann, used to say in his habitually sarcastic 
manner, "Some people seem to think that those characteristics which 
prove most resistant to the attacks of moths and the destructive beetle, 
Anthrenus, must also be regarded as those most resistant to evolutionary 
change." The skulls of all the raptors mentioned above are indeed sur- 
prisingly similar to one another because, to quote Stresemann again, 
"Bones are just like wax in the hands of evolution." 

A. E. Brehm (1890), who as a taxonomist did not otherwise shine at all, 
naively but with justification relied on his own gestalt perception and 
boldly asserted that the harpy was "the hugest of all goshawks" and that 
the hooded eagles (Spizeatus) were "large slender goshawks with rela- 
tively short wings." The correctness of his opinions and the taxonomic 
errors that still persist in the Tierenzyklopadie are both caused by neglect- 
ing the fact, mentioned in One/II/3, that the natural Computer of our 
gestalt perception can take in and evalúate a much greater number of 
data than our rational computation can. Many truths become falsified 
and many obvious facts become invisible if one restricts his methods to 
quantification alone. 

Quantification, however, has the last word in verification, and all that 
our perception can tell us becomes "science" only if and when we suc- 
ceed in confirming it by rational verification. In order to accomplish this 
difficult task with regard to our phylogenetic considerations, we have no 
other method but that of tracing back the way by which our perception 
carne to its results. In other words, we must try to find out, and try to 
describe, what were the single characteristics evaluated by the compu¬ 
tation that our gestalt perception has performed. We are very far indeed 
from regarding perception's function as a miracle, so we can be quite 
certain that all these data must have, in some way, been "fed into" the 
apparatus of perception. As I have already said: if anywhere in biology 
man-made computers are more than a mere model, it is in the physiology 
of perception. 

So it becomes our task to find out which characteristics link the golden 
eagle to the buzzards, the bald eagle to the kites, and the harpy to the 
hawks. These characteristics must be exclusively distinctive of the group 
concerned; they must not be a common possession of the next larger tax¬ 
onomic unit, in our case, of the raptors; if they were, they would not 


IV. The Comparative Method 

support our assumption. Our attempt to compute the relationship 
between the numbers of group-specific characteristics and those of a 
more generalized kind raises the difficult problem of what has to be 
regarded as "one" or a "single" characteristic. There are many character¬ 
istics which superficially seem independent of one another, as parts of 
the same adaptation. Streamlined bodies and sickle-shape fins or wings 
are virtually always part of the same adaptation to swift movement, so 
they ought to be regarded, in our computation of analogous versus 
homologous characteristics, as only one characteristic. The closer to each 
other, on the genealogical tree, the two points are situated from which 
two courses of convergent evolution have taken their departure, the 
smaller is the number of group-specific characteristics in relation to those 
created by the convergent adaptation. With regard to the three groups of 
raptors mentioned above, it is just possible for people with a gift for 
gestalt perception to reach an unequivocal classification. 

There are, however, cases in which the criterion of relative numbers 
fails us completely. There is, for instance, a species of waterfowl in Abys- 
sinia, the blue-winged goose (Cyanochen cyanopterus), which unambigu- 
ously belongs to the subfamily of sheldrakes (Tadornini). Like many other 
waterfowl, these birds have, through adaptation to feeding on grass, 
evolved a goose-like toothed bilí. So have the South American members 
of the genus Cloephaga, which also clearly belong to the Tadornini. Being 
members of the same subfamily, Cyanochen and Chloephaga have so many 
distinctive marks in common, and so few exclusively specific ones, that 
no one can tell whether Cyanochen is a Chloephaga which, through a freak 
of windstorms, was blown over from South America, or whether an 
independent kind of Tadornini has "invented" the grass-eating bilí on 
its own in Abyssinia. No one has yet undertaken an investigation of the 
two genera in detail, searching for Cyanochen-speciñc and Chloephaga- 
specific characteristics. This type of diligent search for distinctive char¬ 
acteristics is the never-ending chore of the phylogeneticist endeavoring 
to disentangle the most recent steps of evolution; it will be discussed in 
detail in Section IV/12. 

11. The Changing Valué of Single Characteristics 

For judging genealogical relationships it is not only necessary to assess 
the relative numbers of homologous and analogous characteristics; one 
must also draw into consideration the rate at which every distinctive 
mark to be used in one's computation is apt to change during the course 
of evolution. The rate at which one and the same characteristic changes 
can be very different in different taxonomic groups. The same distinctive 
mark can be very similar or practically identical in all members of a large 
group, and it can be highly variable in another. The deduction is justified 

11. The Changing Valué of Single Characteristics 


that, during the evolution of the latter group, such a distinctive mark 
tends to change very quickly. Having four legs is a tolerably conservative 
characteristic in tetrapods, yet in one lizard family, the Scincidae, we find 
closely related forms of which some have four legs, some two, and some 
none at all. Among parrots (Psittaci), the coloration and markings of the 
plumage are extremely different, often even in closely allied forms, 
while the form of the bilí is nearly identical throughout. Conversely, in 
the Galápagos finches (Geospizidae) and also in the Hawaiian Drepani- 
dae, the plumage of all genera and species is closely similar, while the 
form of the bilí is subject to extreme variation. Obviously, if one were to 
ascribe to the coloration of the plumage, or to the form of the bilí, a con- 
stant "taxonomic dignity" throughout the class of birds, this would give 
rise to very erroneous conclusions. There is no such thing as a constant 
taxonomic valué for any characteristic. 

In his contribution to Bronn's Klassen und Ordnungen des Tierreiches 
(1891), Hans Gadow has made an interesting thought experiment. He 
chose thirty distinctive marks whose importance was generally recog- 
nized by ornithologists and, ascribing the same valué to each of these 
characteristics, he constructed a taxonomy of birds based exclusively on 
those thirty characteristics. The result was surprising: while it showed 
some striking points of agreement with current taxonomy, it diverged 
on others in the most blatant way. Gadow himself clearly recognized that 
the "intuition" of gifted taxonomists attributed a different weight to the 
same characteristics in different branches of the genealogical tree. He 
was also aware that this "intuition" evaluated a much greater number of 
characteristics than a mere thirty. 

The multiplicity of data fed into and interpreted by our gestalt percep- 
tion enables that great Computer to ascertain the relative speed with 
which a certain distinctive mark is evolving in different groups of ani¬ 
máis, and to attribute a corresponding significance to it with regard to 
every group. The prerequisite for this tremendous accomplishment is, of 
course, that an equally tremendous number of data has been "fed" into 
the Computer, innumerable data concerning each species and also con- 
cerning a huge number of different species. The speed with which a cer¬ 
tain characteristic changes in a certain group can only be computed from 
a comparison with the speed of change in many other characteristics of 
many other groups. The correctness of the end results would only be a 
certainty if the taxonomist could evalúate all characteristics changing 
through evolution in all the members of the group investigated. This is 
obviously utopian, but it is certain that the probability of correct geneal¬ 
ogical interpretation rises with every new characteristic and with every 
new species drawn into consideration—possibly even in a geometric 

Comparative evaluation of the relative changeability of many distinc¬ 
tive marks contributes to furnishing a criterion of homology which cer- 


IV. The Comparative Method 

tainly enters into what is called "taxonomic intuition," although it is con- 
sciously and rationally used only in rare cases. It is a matter of course to 
all taxonomías that the existence of homologous characteristics is a sign 
of phyletic relationship. The inverted conclusión seems banal: of course, 
the likelihood of two characteristics being homologous in two forms of 
life increases with the number of other homologous characteristics both 
possess. This trite statement gains in significance as investigation 
advances to more detailed questions concerning the "microsystematics" 
of a group. The farther this endeavor proceeds, the more the difficulties 
arising from homoiologies increase. In this field, and quite particularly 
in behavior research, it becomes a help that there is a relationship of 
"mutual elucidation" between ascertaining the homology of two single 
characteristics and that of the phyletic relationship between two species. 
As Dagmar Kaltenháuser (1971) has shown in her study of homologous 
motor patterns in dabbling ducks (Anatini), the proof of phyletic rela¬ 
tionship is often less dependent on a demonstration of the homology of 
two motor patterns than, conversely, the proof of the homology of two 
complicated behavior patterns hinges on a demonstration of the relation 
between the two species in question. 

12. The Difficulties and the Importance of 

All our assertions about the phyletic relationship of animáis, extinct or 
extant, are based on a consideration of probabilities. The probability of 
our being correct is in direct proportion, not only to the number of char¬ 
acteristics evaluated, but also to the number of forms of life bearing these 
characteristics. All that has been said in this chapter proves just this. The 
recognition of these facts makes it evident that our reconstruction of the 
great genealogical tree of vertebrates possesses a vastly greater probabil¬ 
ity of being correct with regard to the branching of the oíd, large taxa. 
It becomes equally evident that this probability diminishes in proportion 
to the number of distinguishing marks setting off one group against the 
other. In addition, the documentation by means of fossils leaves us com- 
pletely in the lurch as far as the most recent steps of evolution are 

All this constitutes a severe obstacle to a very important breakthrough 
in modern natural Science, to the synthesis of phylogenetics and 
genetics. It is only the most extensive knowledge of the most subtle 
marks distinguishing a great number of living forms that can help to 
surmount this obstacle. Yet the investigator who concentrates all his 
life's work on the study of one seemingly unimportant group of animáis 
is often regarded as a crank, if not a monomaniac. The man who seems 
inordinately enamored with such a group, counting the bristles on legs 

12. The Difficulties and the Importance of "Microsystematics' 


of different Cladocera, or describing one particular vein in the wing of 
some tiny midge, is all too easily ridiculed, yet it is he who helps to 
accomplish the great breakthrough here under discussion. As Erwin Stre- 
semann tells us, even the great ornithologist Ernst Hartert met with the 
disapproval of his colleagues when he insisted "on the most meticulous 
distinction of local forms" and justified his contention by writing, as 
early as 1899: "Forms that are difficult to distinguish from one another 
must be observed as they occur in nature, and there is nothing in nature 
that we may neglect" [my translation]. One feels tempted to speculate 
that this great systematist may have felt intuitively that a comparison of 
forms that can hardly be distinguished from one another could elucidate 
some of the most essential problems of evolution. 

To the best of my knowledge, Erwin Stresemann was the first to grasp 
this and was the first to take the necessary steps for utilizing the great 
amount of detailed knowledge gathered by ornithologists, as well as the 
huge amount of material accumulated in ornithological collections, for 
the study of evolution. By being a scientia amabilis, ornithology has 
attracted a great number of "amateurs," in the sense defined in One/II/ 
4, with the result that no class of vertebrates has ever become nearly as 
well known, with regard to its systematics, its geographic distribution, 
its ecology, and the breadth of variability occurring within each species, 
as have the birds. Stresemann emphatically opposed the opinión, then 
current among ornithologists, that the continuity of transitions between 
adjacent and very similar species of birds was due to the inheritance of 
characteristics that were caused by climatic or other environmental influ- 
ences. This Opinión was then held by such great scientists as Charles Otis 
Whitman and Bernhard Rensch. As Stresemann demonstrated in his clas- 
sic study on discontinuous variation, a gradual transition between two 
closely allied, yet clearly distinguished species occurs exclusively when 
their habitats are adjacent to each other so that, secondarily, hybridization 
takes place. Stresemann's exploitation of the hoard of ornithological 
material brought a result of the utmost importance in the history of 
biological Science. It proved the mutation theory to be correct and proved 
the assumption of the inheritance of acquired characteristics to be false. 

These new finds opened the way for a new branch of biological Sci¬ 
ence, population genetics, represented in America by Sewall Wright and 
in England by Ronald A. Fisher. Among geneticists, it was Theodosius 
Dobzhansky who most successfully combined the methods of genetics 
with those of phylogenetics. It was an ornithologist, and a pupil of E. 
Stresemann's, Ernst Mayr, who in 1942 wrote Systematics and the Origin of 
Species , a book which really eífected a major breakthrough by bridging 
the gap existing between two hitherto unrelated branches of Science. 
Every student of ethology ought to read this book because, as Stresemann 
said, "... for a long time to come, it will be an unerring guide to the 
systematist, guiding him through the complex maze of phenomena from 
which our predecessors, 150 years ago, vainly sought a way out." 


IV. The Comparative Method 

13. The Origin of Ethology 

Not only this new bridge uniting the study of phylogeny with that of 
genetics has its factual fundaments in the findings of the modest and 
pedestrian search for more and more comparable characteristics in more 
and more related species—a search that was, and still is, looked down 
upon by people lacking an understanding for the great historical facts of 
evolution. Another new branch of biological Science, a side branch that 
sprouted in an altogether unexpected direction, has the same factual 
foundation, and is the subject of this book: comparative ethology. As has 
been explained in Chapter II, the cognitive functions of gestalt percep- 
tion are absolutely indispensable in biological Science. It has also been 
demonstrated that being an "amateur" is, rather surprisingly, the prereq- 
uisite for the full unfolding of these functions. It is easy to understand 
why "amateurism" and gestalt perception play a supremely important 
role in systematics and particularly in the microsystematics discussed in 
Section 10 of this chapter. It is equally easy to understand why ornithol- 
ogy, looked down upon by fools as a scientia amabilis, has given birth to 
population dynamics, to population-genetics and to ethology, three Sci¬ 
ences whose importance is increasingly recognized. The joy experienced 
in the beauty of organisms leads to collecting; the joy of collecting leads 
to systematics; systematics leads to the recognition of the great laws pre- 
vailing in the génesis of the organic world. 

The discovery which may be equated with the origin of ethology is a 
simple one. It was made by microsystematists occupied with their insa- 
tiable hunt for more and ever more comparable characteristics. It was by 
no means a coincidence that the two men who, independently of one 
another, made that discovery. Charles Otis Whitman and Oskar Hein- 
roth, were both ornithologists and both unmistakably true "amateurs," 
both aviculturists and both collectors of live creatures. Entomologists, 
too, enjoy the beauty of organisms, but they tend to collect carefully pre- 
served specimens rather than live ones. Most ornithologists, on the other 
hand, assiduously observe live birds, either in the field or in captivity. 
The passion for collecting assumes an important part as well; Whitman 
kept a collection of doves and pigeons (Columbae) in aviaries; Heinroth 
had at his disposal the rich collection of waterfowl kept at the Berlin Zoo. 
As has been explained in One/II/5, the observation of captive animáis 
has the important advantage of stimulating comparison by showing, side 
by side, different species normally inhabiting far distant countries. 

Under these circumstances a microsystematist on the lookout for com¬ 
parable characters can hardly fail to notice that there are behavior pat- 
terns which represent just as reliable—and often particularly conserva- 
tive—characteristics of species, genera, and even larger taxonomic 
groups, as do any morphological characteristics. In his short paper, "Über 
bestimmte Bewegungsweisen bei Wirbeltieren," (1930) Heinroth clearly 

14. Chapter Summary 


demonstrated that the concept of homology is applicable equally to 
motor patterns and to morphological characteristics. 

As the history of Science shows, it is often necessary for a natural law 
to be revealed in a particularly simple and striking form in order to draw 
the attention of the research worker. Gregor Mendel discovered the 
Mendelian laws when he had the good luck to stumble on their simplest 
possible realization in hybrids differing only with regard to one gene. 
Both Whitman and Heinroth discovered the fact that motor patterns can 
be homologized when they observed the most constant and unchange- 
able sequence of movements existing in the animal world, that is, ritual- 
ized motor patterns of fixed intensity. There are in many animáis motor 
patterns which act as signáis and which, under the selection pressure of 
this function, have evolved in a direction of becoming less ambiguous, 
as will be discussed in Two/II/3. It suffices to say here that motor pat¬ 
terns of fixed intensity lack the degree of variability which, in others, is 
correlated with the variations of specific excitation. Some courtship pat¬ 
terns in pigeons and in waterfowl are typical of this type of signal 

14. Chapter Summary 

Ethology, or the comparative study of behavior, is based on the fact that 
there are mechanisms of behavior which evolve in phylogeny exactly as organs 
do, so that the concept of homology can be applied to them as well as to morpho¬ 
logical structures. As all behavior is based on the function of structures 
which have phylogenetically evolved, the statement of this fact is (or 
ought to be) redundant. Nevertheless, it was left to two zoologists to dis- 
cover. They did so while searching for distinctive marks that could be 
used in the detailed systematics of the group they were investigating. 

When used by biologists, the verb "comparing" means using the sim- 
ilarities and dissimilarities of extant forms of life in an attempt to recon- 
struct their descent from a common ancestor. Such a reconstruction can 
be accomplished by means of a diagram into which no hypothesis needs 
to enter. If one symbolizes, by the use of vertical lines, the species of 
animáis living today, and subsequently connects them by horizontal 
lines representing characteristics common to a number of them, all that 
is necessary to give the resultant diagram the form of a tree is to order 
the symbols of common distinguishing marks in such a manner that 
those which connect the greatest number of living species are marked at 
the lowest level (at the bottom of the diagram), and to arrange the others 
in a sequence corresponding to the width of their distribution among 
animal forms. This has been done in the diagram of Figure 6, using the 
chordates as an example. 

Practically everywhere in nature that we perceive the well-known 


IV. The Comparative Method 

form of a tree, the assumption proves correct that what is found has 
grown, in other words, that the parts now representing the tree top are 
youngest and that, at the first stages of growth, the whole tree was rep- 
resented by what now forms the lowest part of the trunk. Even without 
the documentaron stemming from other sources, such as paleontology 
and animal geography, the data brought to light by comparison alone 
would render the hypothesis of common descent a near certainty. 

Complete certainty is attained by the testimony of paleontology. The 
fossils characterized by unmistakable distinguishing marks as being 
ancestral are found in the lowest strata of the earth, and in the resultant 
sequence of the earth's layers we find fossil organisms in exactly the 
same sequence as that which has to be postulated on the basis of a com¬ 
parison. In the diagram of Figure 6, the geological epochs can be repre- 
sented by horizontal lines which coincide, without constraint or the least 
contradiction, with the conclusions drawn from comparisons concerning 
the sequence in which new characteristics first made their appearance. 

Organs and characteristics which owe their similarities to descent from 
a common ancestor are called homologous. There is one other way by 
which similarities can come to be: two or more forms of life, which need 
not be related to each other, can evolve organs amazingly similar in 
almost every detail by becoming adapted to the same function. The clas- 
sical example for this process of convergent evolution is represented by 
the camera eyes independently evolved by vertebrates and cephalopods 
(Octopus). Similarities arising from convergent adaptation are called 

For correctly assessing genealogical relationships, it is of the utmost 
importance to distinguish clearly between homologies and analogies. A 
consideration of probabilities is the surest way to this end. A porpoise 
(Delphinus) and an ichthyosaurian are strikingly similar to each other 
with regard to the streamline forms of their bodies and to the organiza¬ 
ron of their flippers, but these similarities are few in number while, on 
the other hand, hundreds of distinguishing marks prove the ichthyosau- 
rus to be a reptile and the porpoise to be a mammal. Any attempt to 
explain the similarities between the two by the assumption of common 
descent would have to explain the innumerable mammalian character¬ 
istics of the porpoise and the reptilian characteristics of the icthyosaurian 
on the basis of convergent adaptation—which would obviously be 

The nearer to each other, on the genealogical tree, the two starting 
points of convergent adaptation are situated, the less confidence can be 
placed in this calculation of probability. For example, the marks distin¬ 
guishing a small group of raptors from each other, such as the hawk-like 
(Accipitrini) and the kite-like (Milvini), do not appreciably exceed in 
number those marks which both groups have evolved through conver¬ 
gent adaptation when each developed so-called eagles. The fewer the 

14. Chapter Summary 


characteristics distinguishing two groups, the more difficult it becomes 
to sepárate homologies from analogies. This is why we know more about 
the older and greater divisions of the genealogical tree than we do about 
the smaller and more recent ones, although the latter are much more 
interesting to those concerned with population dynamics and with 

In order to acquire more insight into the most recent events of phy- 
logeny, the microsystematist is forever searching for more and more 
comparable characteristics. The results of his patient search made possi- 
ble the important breakthrough on which the unification of phylogeny, 
population dynamics, and genetics is exclusively based. An unexpected 
and additional result of this search is ethology, that is, the comparative 
study of behavior, which also originated through microsystematics. In 
the diligent search for comparable characteristics, Whitman and Hein- 
roth hit upon the fact that behavior patterns could be just such charac¬ 
teristics. Thus the comparative study of behavior, called ethology, has 
arisen from systematics as an unexpected new branch of Science. 

The discovery of behavior patterns which, in phylogeny, "behave" 
exactly as morphological characteristics do, is of the greatest importance 
for one particular reason: it proves beyond a reasonable doubt, and on 
the basis of all the mass of evidence proving evolution as such, that fixed 
motor patterns, comparable in their minutest detail, owe their special 
forms to genetic programs that carne to be such as they are through the 
processes of evolution in exactly the same manner as did the programs 
for all morphological structures. 

Part Two 

Genetically Programmed 


Chapter I 

The Centrally Coordinated 
Movement or Fixed Motor Pattern 

1. History of the Concept 

To give an exact definition of what we mean when we speak of a fixed 
motor pattern or a fixed action pattern is diíficult because one should not 
inelude any working hypotheses in the definition of any biological func- 
tion or structure that has not already been thoroughly analyzed. Still, it 
is possible to confess that we believe that the hard core of what we cali 
fixed motor patterns consists in centrally coordinated sequences of 
endogenously generated impulses, and that this coordination has 
evolved phylogenetically and is very resistant to any individual modi- 
fication. A fixed motor pattern's most important distinction from motor 
patterns that are not fixed and, simultaneously, the cogent argument for 
its being genetically programmed, consists in its taxonomic distribution. 
Fixed motor patterns show, from species to species and from genus to 
genus, similarities and dissimilarities that concur strictly with those of 
morphological characters. The discovery of these faets of similarity and 
dissimilarity is due to the work done by C. H. Whitman and O. Heinroth. 
They discovered that the concept of homology could be applied to motor 
patterns in certain "ritualized" movements of courtship in pigeons (Col- 
umbidae) and in waterfowl (Anatidae). It is typical for discoveries con- 
stituting major breakthroughs that they are made with objeets in which 
the newly found laws of nature are represented in the simplest possible 
ways. The elassie example of this, as already mentioned, is Gregor Men- 
del's hitting upon monohybrids, that is, hybrids between races of plants 
which differ in only one gene. The courtship movements in question also 
obey the simplest possible laws. They obey the so-called all-or-nothing 
law of nervous discharge; in other words, their identity is not veiled by 


I. Fixed Motor Pattern 

the great variability of the forms in which other motor patterns appear 
in correlation with the varying intensity of their specific excitation. As 
will be discussed in the following section, even a low excitation can elicit 
barely noticeable "intention movements," and with rising excitation, a 
gradual scale of transitions leads from this "intention" to the fully inten- 
sified motor pattern achieving its teleonomic function. In some particular 
motor patterns which function as signáis , this intensity-correlated varia¬ 
bility has been abolished in the interest of unambiguity: when they are 
performed at all, it is always with the same "typical intensity," as Des- 
mond Morris (1957), who discovered the phenomenon, has described it. 

These kinds of motor patterns of courtship, always performed monot- 
onously and with the same intensity and, furthermore, hardly overlaid 
and veiled by simultaneous orientation responses, were the object of 
study through which Whitman and Heinroth discovered, independently 
of one another, first, that they were highly chavaderistic for each species, 
and second, that their similarities and dissimilarities from species to spe¬ 
cies, from genus to genus and, indeed, from one taxonomic group to 
another, were most exactly correlated to the similarities and dissimilari¬ 
ties of all morphological characters. This meant that the concept of 
homology could be applied to these behavior patterns just as well as to 
morphological properties. Historically, as has been explained in the 
introduction, it was the search for characteristics classifiable as homolo- 
gous that led Whitman and Heinroth to their truly epoch-making 

Neither of them ever formed a hypothesis concerning the physiolog- 
ical nature of their discovery. It is, therefore, worth extensive consider¬ 
aron that, in their endeavor to collect motor patterns that could be used 
as characters in taxonomy, they unequivocally chose processes of one sin¬ 
gle physiological type. This chapter and most of the second part of this 
book will be dedicated to this particular physiological process—the 
endogenous, centrally coordinated movement, which is also called the 
fixed motor pattern. 

Being not particularly interested in the physiological aspects of the 
process, Heinroth defined it exclusively by the function of the whole 
program. Thus this concept embraces what we now know to be three 
physiologically diíferent functions: first, the "drive" to search for a cer- 
tain stimulus situation; second, the selective response to it-its "innate rec- 
ognition"; and third, the discharge of an equally innate motor activity 
coping with the situation. Behavior systems integrating these three parts 
are indeed very common in nature, which explains and almost excuses 
Heinroth, as well as myself, for the length of time we believed that the 
arteigene Triebhandlung (species-characteristic drive action) was the most 
important, if not the only element of which all animal behavior is built 
up. At a time when the reflex was generally thought to be the most 
important, if not the only element of behavior, it is not surprising that 

1. History of the Concept 


the reflex-like response to a certain stimulus situation was not concep- 
tually separated from the equally specific motor activity that followed. 

The scientists who at that time were investigating orientation 
responses tended to formúlate similar, all-embracing concepts. Even 
when Alfred Kühn, in his classic book, Die Orientierung der Tiere im Raum, 
spoke of a positive or negative phototaxis, his conceptualization of this 
term included not only the organism's turning in space toward or away 
from the source of light, but also the locomotion in those directions. 

The combination of the unlearned "knowledge" of an object with an 
unlearned "skill" for coping with it is striking to the observer when, for 
instance, a hand-reared, totally inexperienced young goshawk catches a 
pheasant flying across the room in midair and alights on the córner of a 
wardrobe with the prey already dead in its claws. Heinroth writes: "This 
first 'professional' act of the hawk made an unforgettable impression 
upon us." 

Frequent as the functional unity of a releasing receptor mechanism 
and an innate motor pattern is, it is far from being the only way in which 
these two elements can collaborate in organized behavior. For this reason 
it is not quite correct when, in Bullock's new textbook on neurophysi- 
ology (1977), it is counted as a constitutive characteristic of the innate 
motor pattern that it can be selectively released by certain specific com- 
binations of external stimuli. There are other ways in which these ele¬ 
ments can be put together in organized systems of behavior, one of 
which has been analyzed by Prechtl and Schleidt (1951). In some young 
mammals a motor pattern serving the search for the mother's teat is 
continuously at work as long as the baby is awake and is inhibited 
only when it has found the teat, as will be discussed in detail in 

The innate ability to "recognize" a biologically relevant stimulus sit¬ 
uation on one hand and, on the other, the innate "skill" to deal with it 
in the teleonomically correct way, are based on two physiological orga- 
nizations which are entirely different from each other. As has been men- 
tioned in the Introductory History, Charlotte Kogon was the first to grasp 
the necessity of further analyzing Heinroth's concept of the arteigene 
Triebhandlung. In 1941 she wrote: "In consistent continuation of Lorenz's 
terminology, it is not possible any longer to speak of an instinctive 
'action', but only of instinctive movement, a consequence which, how- 
ever, he fails to draw." As has been mentioned, the conclusión was 
indeed drawn immediately afterwards. The dynamics of the interaction 
between the "instinctive movement"—now termed fixed motor pattern in 
English—and the innate releasing mechanisms were fully understood 
only after the experimental analysis accomplished by Seitz, as will be 
discussed in Section 14 of this chapter. 

A fundamental prerequisite for the success of this analysis was a really 
thorough familiarity with the different forms in which a motor pattern 


I. Fixed Motor Pattern 

can appear at different levels of specific excitation. This knowledge could 
never have been acquired by any other methods than those described in 
the second chapter of Part One. 

2. Differences in Intensity 

The so-called "all-or-nothing" law, so well known as a part of early neu- 
rophysiology, does not hold true with regard to instinctive movements, 
except for those movements performed with typical intensity, as 
described on page 108. Quite the contrary, when the specific excitation 
activating a certain fixed motor pattern begins to rise even by minimal 
degrees, this finds its expression in movements: slight indications, 
"hints" of the motor pattern appear. Although their amplitude is so small 
as to be hardly discernible, the "impulse melody," as von Holst called it, 
becomes recognizable to anyone familiar with the pattern. These phe- 
nomena reveal what actions are to be expected from the organism in the 
near future; in other words, they indicate the organism's "intentions," 
and for this reason Heinroth called them intention movements. A hawk 
in which the specific excitation to take flight begins to awaken, performs 
aiming movements with its head and neck, treads alternately on one foot 
and then the other, crouches in preparation for jumping off, and even 
half opens its wings. When in a night heron the specific excitation for 
nest-building begins to well up, it will assume the attitude of standing 
in a nest, while, in fact, still standing on a branch or on the ground. Its 
shoulders are lowered and its tail is raised; it fixates for a moment on a 
dead stick or some other potential building material, only to relapse in 
the next second into its former, relaxed sitting position. The faculty of 
recognizing these kinds of hints or indications is indispensable to the 
ethologist trying to predict behavior. 

An increase of action-specific excitation, first noticeable in the inten¬ 
tion movements, can level off at any valué of intensity already attained, 
or it may begin to wane again at any given moment. A greylag goose, in 
which excitation for taking wing has stopped just short of the threshold 
for actually doing so, can come to a full stop in a deep crouch preparatory 
to jumping with its wings half extened—a position that makes it look 
like a badly stuffed bird—only to stand up again and carry on with what- 
ever it had been doing before it had begun to want to fly. 

Still, if one is sufficiently familiar with the series of different forms 
that the goose's preflight movements take at different degrees of mount- 
ing excitation, one is perfectly able to foretell, and with tolerable exact- 
itude, whether or not the bird will finally fly. I can always impress visi- 
tors by saying: "Look at that goose. It thinks it is going to fly, but it 
won't," and have my predictions come true. And I could do this long 
before strenuous self-observation revealed to me which were the data I 
was using to derive that information: these were the speed with which 

2. Differences in Intensity 


the goose passed from one State of intensity to the next higher one. If 
the bird proceeds slowly through stage one (raising its neck and uttering 
preflight calis) and then gets stuck at stage two (which is bill-shaking) 
for a long time, often then relapsing to stage one (calling without bill- 
shaking), it is certain that the intensity will not rise to stage three (mov- 
ing the wings out from under the flank feathers) ñor to stage four 
(crouching for the jump), and least of all to the stage of jumping into the 
air and taking wing. If stages one, two, and three are passed in quick 
succession, one can be sure that the curve of mounting excitation will 
not flatten out suddenly and that the following stages are going to be 
reached. The waxing as well as the waning of excitation seems to contain 
a sort of inertia within itself; the curve which excitation describes cannot 
break oíf sharply and can, therefore be safely extrapolated so that, after 
having registered a portion of its course, the actions which are to follow 
can be predicted. 

Quite another question concerns the criteria by which an observer can 
recognize a fixed motor pattern when all he can see of it are but slight 
hints, which are very diíferent from its full performance. The answer to 
this question has been given by Schleidt in his paper, "How 'Fixed' is 
the Fixed Action Pattern?" (1974), which will be discussed in more detail 
in Section 14 of this chapter. What remains constant during the whole 
series of transitions leading from the slight intention movement to the 
full performance of an innate motor pattern are first, the relations 
between the phases of the elementary movements of the pattern and sec- 
ond, the equally unchangeable proportions between the amplitudes of 
these movements. By means of these characters the gestalt perception of 
the observer recognizes the "impulse melody"—as von Holst has called 
it—of the behavior pattern, just as a melody can be recognized even if 
only parts of it can be heard and it is played so softly that it is hardly 

Erich von Holst demonstrated that the diíference in the forms which 
a fixed motor pattern takes at diíferent intensities is to be explained in 
the following manner. The automatic cells generate the impulses which 
are coordinated by central coordination, as will be described in Section 
13 of this chapter. These impulses activate motor cells in the anterior 
horn of the spinal cord, and these cells respond with slightly diíferent 
thresholds to the same quality of excitation. At a low intensity, only a 
fraction of the motor cells are responding, and their number increases 
with rising excitation "by recruitment," as von Holst has termed it. The 
margin between the highest and the lowest thresholds of the motor cells 
involved diífers with diíferent motor patterns. For those patterns serving 
as signáis, it is in the interest of unambiguity that their form should be 
held as constant as possible, and this is achieved by narrowing the mar¬ 
gin between the thresholds of motor cells, so that even with changing 
intensities the motor pattern can vary only within narrow limits. For 
such instances Desmond Morris (1957) speaks of a "typical intensity" or. 


I. Fixed Motor Pattern 

when the patterns actually obey the all-or-nothing law, of a "fixed inten- 
sity." As has been mentioned, motor patterns of courtship, which are the 
paradigms of signáis as well as of fixed intensities, have become impor- 
tant in the history of our Science because it was through them that Whit- 
man and Heinroth discovered that motor patterns could be acknowl- 
edged as homologous (One/I/1). 

3. Qualitatively Identical Excitation Activating Different 
Motor Patterns 

The phenomenon of recruitment is not always sufficient to explain the 
differences in the movements that are successively activated by the esca- 
lation of one specific excitation. There are many cases known in which 
an excitation of obviously identical quality calis forth, by its gradual 
increase, a series of motor patterns which, though unambiguously cor- 
related to that same kind of excitation, are quite different from each other 
in the form of their movements. If the specific excitation of fighting is 
elicited in a male of the genus Haplochromis, a cichlid that was one of the 
main objects of Seitz's investigations (1940, 1941), the fish first responds 
by spreading its median fins and assuming "nuptial" (more correctly, 
"displaying") coloration. Then the fish approaches the releasing object— 
a dummy or a live rival—and orients in such a way that the largest con- 
tour of its body is presented at a right angle to the adversary's line of 
sight. This is called a broadside display. In this position, the fish extends 
the gilí flap or branchiostegal membrane downward so that it forms a 
broad, black crest enlarging the fish's contours when seen from the side. 
After this, with mounting excitation, there follows a simultaneous con- 
traction of all the muscle segments on the side nearest the enemy. This 
synchronous movement is completely different from the metachronous 
action of the fish's muscle segments, the undulating movement of swim- 
ming forwards. The simultaneous contraction does not result in loco- 
motion but in an extremely strong, sideways sweep of the widely-spread 
tail fin in the direction of the rival. This movement causes an abrupt and 
strong pressure wave which is certainly perceptible to the other fish 
through the organs of the lateral line, and perhaps conveys a measure of 
the antagonist's power. Up to this stage the changes in the displaying 
fish's motor patterns can be explained by recruitment, because all the 
muscle activities which occur at the lowest steps of mounting excitation 
have continued; none has been "switched off." Now, however, with 
excitation rising still higher, the fish suddenly changes its orientation 
and relinquishes its stiff, broadside display in order to bite or, more cor¬ 
rectly, to ram. The mouth is protruded so that the teeth point forward, 
and these are thrust forcefully against the antagonist toward which the 
fish has necessarily had to turn. Frequently both combatants turn simul- 
taneously so that mouth is thrust against mouth. This is accompanied in 

4. Unity of Motivation 


many cichlids by a frontal display during which the gilí membrane 
assumes a position different from the one shown during the broadside 
display: the membrane is extended at a right angle to the longitudinal 
axis, increasing the body contour when seen from the front. This phase 
is very short and even rudimentary in Haplochromis, while in other cich¬ 
lids it has become ritualized into an elabórate pattern of mouth fighting. 
In Haplochromis , the adversaries almost immediately change orientation 
so as to be able to ram one another on the unprotected flank. From this 
moment on the real, damaging battle is joined; the fish swim around each 
other in narrow circles, a movement which has been called the "carou- 
sel" by Seitz. The ramming thrusts find their mark again and again; fins 
are torn and soon scales are floating around. 

Under normal conditions these motor patterns follow one another pre- 
cisely and in the sequence just described. Only when an abnormally high 
degree of excitability prevails, for instance, after very long isolation, does 
it happen that, when presented with a releasing stimulus situation, the 
fish might show the described movements out of their regular sequence, 
that is to say, the fish might skip the movements correlated to low intens- 
ities of excitation and proceed, almost from the very beginning, to tail 
beating or biting. Except for these very rare occurrences, one might say 
that the fish is unable to show its lateral display before having assumed 
nuptial color, ñor to deliver a tail beat before having oriented broadside 
to the antagonist, and so on. The motor patterns just mentioned are con- 
nected by gliding transitions and it seems safe to assume that their inten- 
sity-correlated differences can be explained on the basis of recruitment 
of motor cells. For the same reason it is equally safe to assume that they 
are the expression of one kind of excitation, different only in quantity 
and not in quality. 

4. Unity of Motivation 

In the cases of motor patterns not grading imperceptibly into one 
another but separated by sudden cessations of certain movements and 
equally sudden emergences of others, as are tail beating and biting, this 
is a problem of whether excitation changes in quality or only in quantity. 
To the naive observer, it would seem obvious that the latter were true, 
but this intuitive assumption must be supported by rational confirma- 
tion. The first verification consists of the absolute predictability of the 
subsequent stage of intensity—once the preceding stage has been gone 
through. When Seitz demonstrated cichlid fighting to our students in 
Konigsberg, he accompanied the actions of the fish with a running com- 
mentary, in much the same way a radio announcer would do for a boxing 
match. But just as he was wont to do when showing one of his own beau- 
tiful films, he made allowance for the reaction times of his listeners by 
anticipating, by a few seconds, what the fish were going to do. Often he 


I. Fixed Motor Pattern 

had already mentioned the tail beat, and had to stop a second or so, wait- 
ing for the fish to do what he had just said they would do. It was remark- 
able how few students noticed this and were surprised by his prescience. 

In addition to the possibility of reliably predicting the movements cor- 
related to the next, higher stage of intensity by means of observing a 
precedent step, another argument for assuming one qualitatively con- 
stant excitation is the frequent occurrence of mixtures, and of extremely 
quick switches from one pattern to the next and back again. Also, gen- 
uine intention movements associated with the next higher stage of inten¬ 
sity can be observed while lower intensity patterns are still being per- 
formed. In a Haplochromis male, the first hints of turning towards an 
adversary while still in the phase of broadside display are just as much 
intention movements as are the first beginnings of preflight movements 
in a quiescent heron. 

A third argument for assuming that it is a qualitatively identical exci¬ 
tation activating the graded scale of different intensities lies in the com¬ 
plete lack of "inertia" shown by the transitions from one motor pattern 
to the next. As will be discussed in One/V/4, an animal needs a consid¬ 
erable amount of time to change its "mood" (Stimmung, Heinroth), that 
is to say, to switch from one system of instinctive behavior to another. 
The required period is that much longer the more complex and inte- 
grated the two systems are and, also, the stronger the inhibitive influ- 
ence they exert on one other. There is one exception to this rule: as will 
be explained in Two/VIII/2, the so-called máximum selecting system can 
effect lightning-like switches between two competing motivations, for 
instance, between attack and flight. Otherwise, a quick transition from 
one motor pattern to another argües for the assumption that one moti- 
vation is activating both. 

A fourth argument for the unity of excitation is the fact that one and 
the same releasing mechanism, in other words, the same stimulus con- 
figuration, elicits all of the motor patterns correlated to different intens¬ 
ities. This is in perfect agreement with the fact found by Erich von Holst: 
typical series of motor patterns, gradated in correlation to intensities of 
excitation, can be elicited from the same locus by electrical stimulation 
of gradated strength, and in exactly the same sequence in which they 
appear in an intact animal reacting to external influences of appropriate 
stimulus configurations. 

Lastly, the strongest argument for assuming a common physiological 
factor is this: The organism's readiness to perform any one of a set of 
innate motor patterns fluctuates parallel to its readiness to perform any 
other. If an activity correlated to a low intensity of excitation can be elic¬ 
ited with very slight stimulation, it can be predicted with absolute cer- 
tainty that those correlated to the higher grades of intensity will be 
equally facilitated. If such a "weak" motor pattern can be elicited only 
by strong stimulation, it becomes predictable that the high intensity pat- 

5. The Method of Dual Quantification 


tern will prove impossible to elicit at all. These parallel fluctuations of 
threshold could not have been discovered without simultaneously dis- 
covering the laws prevailing in the function of the releasing mechanism. 
Both discoveries are due to research done by Alfred Seitz (1940, 1941). 

5. The Method of Dual Quantification 

In the preceding section I was forced to treat some facts as if they were 
already known, although they really become clear only through the 
investigations conducted by Seitz, which now will be described. In this 
description and for similar reasons, I shall be forced to anticipate some 
discoveries that will not be dealt with until the next chapter. Analogous 
difficulties are present in the analysis as well as in the didactic represen- 
tations of any complex system in which each part is interacting with 
every other part. The methods of coping with these difficulties have 
already been explained in One/II/2. 

Before we had grasped the laws prevailing in the imperceptible gra- 
dations of the forms which motor patterns assume in correlation with 
the gradated intensities of one and the same quality of specific excitation, 
it was, in principie, impossible to discover the fact that quite a number 
of stimulus configurations exert a qualiatively identical but quantita- 
tively different releasing influence. An inversión of this sentence is 
equally true. In other words, the laws prevailing in the different intens¬ 
ities of specific excitation and those governing the function of innate 
releasing mechanisms could only be discovered simultaneously. This is 
exactly what Seitz did. His work can be regarded as a model of analysis, 
doing justice to the systemic character of its object. 

The primary and naive aim of Seitz's undertaking was to discover the 
quantitative relationship between the impinging stimulus and the motor 
activity elicited. His subject was the fighting behavior of a Haplochromis 
species then called Astatotilapia strigigena. Male fish of this species 
respond easily to rather simple dummies. The first and, at that time, dis- 
appointing result was that any dummy presented to the fish did not elicit 
the same response even twice in immediate succession; during the sec- 
ond presentation the response was already perceptibly weaker, and still 
weaker during the next. In such a situation the experimenter quite 
unconsciously tries to increase the eífectiveness of the dummy by the 
simple means of waggling it a bit—and behold! the response regains 
some of its former intensity! 

This simple and unintended result may be regarded as the basis of 
Seitz's further procedure: he grasped the essential fact that, in any 
dummy experiment, he had to deal with two variables. First, with the 
internal readiness of the animal to perform a certain motor pattern, a 
readiness which obviously decreased with every performance; and, sec- 


I. Fixed Motor Pattern 

ond, with the varying efficacy of different stimulus situations, the wag- 
gling dummy being obviously more effective than the motionless one. It 
also became clear that the same activity could be elicited by a strong inter¬ 
nal readiness and a weak external stimulation and vice versa. All of the 
subsequent experiments conducted by Seitz were already based on the 
hypothesis that one sort of action-specific excitation appearing in one set 
of intensity-gradated motor pattern was released by one receptor mech- 
anism responding to a specific configuration of key stimuli. On the basis 
of this hypothesis he devised the elegant method of dual quantification. 

To find the degree of effectiveness inherent in a dummy, he presented 
it to the fish and recorded the strength of the response. Then he con- 
fronted the experimental animal with the strongest stimulation at his 
disposal, that is to say, another fish he had first induced to assume full 
fighting coloration, and again recorded the strength of the response. In 
this way he determined the decrease of readiness caused by the presen¬ 
taron of the dummy; in other words, he ascertained how much readiness 
was "still left" after the first part of the experiment, the presentation of 
the dummy. The greater the effect of the dummy, the weaker was the 
subsequent response to the máximum stimulation. Repeating this exper¬ 
iment with a series of dummies, Seitz was able to determine the differ- 
ences between the effects of the several dummies independently of the 
State of excitability in which the experimental animal happened to be at 
the time of each experiment. 

Seitz's investigations of the varied effectiveness of different dummies 
showed, as an unexpected result, that the similarity between the dummy 
and a real fighting male was irrelevant to the releasing effect, at least as 
similarity appears to our human perception. My own former term for the 
perceptual mechanism releasing instinctive behavior had been angebor- 
enes auslósendes Schema —innate releasing scheme. The word Schema — 
scheme—suggests an image which, although much simplified, still con- 
veys an overall idea of the whole unit that it represents. As early as 1935, 
I had pointed out that the innate recognition of any fellow member of 
the species did not depend on perceiving and recognizing this other 
animal as a unitary whole. Quite the contrary, each of the stimulus con- 
figurations which were effective as "key stimuli," evoking a certain 
behavior pattern, acted independently of each other; a tuft of the red 
breast feathers evokes the fighting response of a robin redbreast in a 
manner qualitatively indistinguishable from the bird's reaction to a tape 
recording of its own song. Still it was surprising to find that the reaction 
of Haplochromis males to the dummies which elicited rival fighting did 
not depend at all on their similarities to the real rival. An elongated 
block of grey plasticine attached to a glass rod holder evoked intense 
fighting if it were moved in such a way as to imitate the tail beat of a 
rival—an easy thing to accomplish by simply rotating the glass rod 
between the fingers. Even if this simple dummy were oriented so as to 

5. The Method of Dual Quantification 


be parallel and cióse to the fish, as a rival would be during a broadside 
display, it evoked a fighting response, although of less intensity than the 
response to the imitation of the tail beat. Seitz conducted a series of 
experiments in each of which another character of the conspecific rival 
was tested separately, and its releasing valué determined by the method 
of dual quantification. The single characters, such as the blue gloss of the 
flank, the black oblique fine across the eye, the black crest below the 
head imitating the extended gilí membrane, the black margin along the 
upper edges of the median fins and the reddish margin below, each had 
its own releasing effect when represented on a dummy that otherwise 
did not resemble a rival male at all. Signal movements, on the whole, 
proved more efficacious than characters of form and color. The effect of 
orienting the dummy parallel to the fish was proved to be greater the 
larger the vertical surface presented was, but if the surface were too 
large, the male reacted by fleeing, as it would normally do when con- 
fronted by a much larger rival. The effect of imitated tail-beating proved 
stronger than that of any single morphological character, and was infe¬ 
rior only to the effect of a ramming thrust that could easily be simulated 
by prodding the fish's side with the glass rod. This usually provoked an 
instantaneous counterthrust by the experimental animal. 

In another series of experiments, Seitz investigated the cumulative 
effects of stimuli whose efficacy he had previously ascertained. Always 
maintaining the method of dual quantification, that is to say, compen- 
sating constantly for changes in the fish's internal readiness, Seitz was 
able to compare the releasing effects of different combinations of stimuli 
and to formúlate some very interesting equations. For example, the effect 
of all morphological signáis plus parallel orientation was the same as that 
of a grey square dummy performing the tail beat. Through patient exper- 
imentation of this sort, Seitz was able to formúlate the following highly 
important conclusión: Any character possessing a releasing valué in 
itself, retains that same valué in any combination with any other char¬ 
acter releasing the same activity. The releasing effect of the dummy is 
dependent on the sum of the releasing characters which it embodies. This 
is now called the lazv of heterogeneous summation. The validity of this law 
was later convincingly confirmed by Heiligenberg (1963) and Leong 
(1969) who used an entirely different method. Most importantly, they 
could show that the relationship between key stimuli was truly additive 
and not multiplicativo, as stimulus effects often are. They also showed 
that Seitz was justified, on the basis of his data, to exelude the multipli¬ 
cativo effect of the stimuli investigated. 

In cannot be sufficiently emphasized that the quantification of stimu¬ 
lus effects was only possible on the basis of a very complete knowledge 
of the particular form which a motor pattern assumes at a certain inten¬ 
sity of specific excitation. This particular form furnished the scale by 
which the effect of stimulation was measured. 


I. Fixed Motor Pattern 

6. Action-Specific Fatigue 

All that has been said concerning the law of heterogeneous summation, 
and concerning the constant releasing valué of the stimulus configura- 
tions involved, proves clearly how reliably one can infer, from one sin- 
gle response to a stimulus of known efficacy, the animal's present State 
of readiness to perform a certain action pattern. The same intensity of 
motor response can be kept constant throughout a number of perfor¬ 
mances if one increases the strength of external stimulation so as to com¬ 
pénsate for the decrease of internal readiness. This is demonstrated 
when, for instance, at the first presentation of a square of grey plasticine 
a fish has reacted with a broadside display and tail beating and, at the 
second presentation, it only reached a mere broadside orientation, but 
on being confronted at a third trial with a similar model with blue gloss 
added, the fish regained the intensity shown during the first presenta¬ 
tion. By adding further stimulus configurations the action-specific exci- 
tation could be kept at the same level for even a few more triáis. 

On the receptor side, the decrease of the internal readiness to perform 
a certain motor pattern finds its expression also in an increase in the 
threshold valúes of releasing stimuli. Both phenomena together may be 
described as action-specific fatigue or exhaustion. That which is fatigued 
or exhausted corresponds exactly to the entity called species-character- 
istic drive action, in other words, to a system consisting of appetitive 
behavior, an innate releasing mechanism and a consummatory instinc- 
tive motor pattern. As we already know, each of these constituents is the 
function of an entirely different physiological apparatus and, therefore, 
we are confronted with a problem: Which of them is exhausted? In 
which of them is the process of fatigue situated? 

Action-specific fatigue or exhaustion has one character in common 
with a process called habituation (which will be discussed later in Three/ 
II/2). Both action-specific exhaustion and habituation concern only one 
specific activity and the equally specific stimulus situation which releases 
it. This parallel could lead to thinking that the whole process of fatigue 
takes place on the receptor side alone. Seitz and I, at the time of his classic 
experiments, tended to believe the opposite: our impression—influenced 
perhaps by the hypothesis held by Erich von Holst—was rather that it 
was on the side of the motor activity that "something" was used up, 
which determined the degree of excitability. Von Holst suggested a spe¬ 
cific neurohormone which was consumed or eliminated by the perfor¬ 
mance of the act. As shall be explained soon, this simplistic assumption 
was not quite correct, but it was legitímate and heuristically valuable 
insofar as the fatigue of the whole complicated afferent system, which 
comprises everything from the peripheral sensory cells to the highest 
levels of the central nervous system, did not play a decisive role in the 
processes we were investigating. 

6. Action-Specific Fatigue 


Our assumption, that some sort of motor excitability was accumulated 
while the motor pattern was not performed and then used up by its per¬ 
formance, was supported by one important observation: When the readi- 
ness to perform a certain motor pattern has been exhausted as completely 
as possible by the use of the strongest stimulation attainable, it takes 
some time for the specific excitability and the releasing thresholds to 
reach their "normal," that is to say, their average valúes. If, however, 
stimulation is withheld still further, neither the rising of excitability ñor 
the concomitant lowering of thresholds will stop at any certain prede- 
termined "normal" valué, but will go on until the animal is ready to 
respond to extremely slight stimulation, that is to say, to stimuli which 
are very far from characterizing the biologically adequate situation in 
which the activity would accomplish its teleonomic function. Lissman 
studied the effect of this phenomenon on the agonistic movements of the 
Siamese fighting fish (Betta splendens) and drew the correct inferences 
through measuring the process of threshold lowering which will be dis- 
cussed further in the next section. 

Seitz and I assumed that these regular fluctuations of internal readi- 
ness could be explained on the basis of a "damming up" of a specific 
excitability, which is subsequently "consumed" during the motor activ¬ 
ity. This explanation seemed convincing because of some results 
obtained with spinal fish in which an analogous phenomenon, the so- 
called spinal contrast, was observed, in which no specific receptor orga- 
nizations were involved. This will be described in Sections 12 and 13 of 
this chapter. The lawful fluctuations of readiness for a certain instinctive 
motor pattern are mainly dependent on processes taking place in the 
motoric sector of the system, although there are very similar ones which 
concern the receptor side. It was rather fortúnate that we did not know 
about the latter at the time of Seitz's experiments on dual quantification; 
they can be neglected in the fighting of Haplochromis without constitut- 
ing a serious source of error. 

When H. Prechtl later (1949) investigated the receptor organization 
releasing the gaping movements of young chaffinches (Fringilla coelebs), 
he demonstrated that a number of stimulus configurations combine their 
several releasing effects according to the role of heterogeneous summa- 
tion. They consisted of a) a slight percussing of the nest, b) a very high 
chirping tone, and c) the optical effect of the parent bird—or, for that 
matter, the hand of the human experimenter—approaching the nest 
from above. Furthermore, gaping can be released by a slight tapping on 
the nestling's head. When Prechtl presented one of these stimuli until 
the response to it had ceased completely, any one of the other stimuli 
would still be able to evoke gaping. The longest series of gaping move¬ 
ments was obtained when any one of the effective stimulus combinations 
was exchanged for another one just before the effect of a prior combi- 
nation had expired. The general effect was exactly analogous to that 


I. Fixed Motor Pattern 

which an experienced chef strives to achieve by means of a cleverly 
devised sequence of dishes. Heterogeneous summation was clearly dem- 
onstrated by simultaneously presenting two or more of the aforemen- 
tioned stimulus configurations. Prechtl's experiments showed that not 
only processes of fatigue on the motor side, but equally those on the 
receptor side of a species-characteristic drive action could be responsible 
for changes in the organism's readiness to perform. 

It was Ludwig Franzisket (1953) who succeeded in analyzing and dif- 
ferentially quantifying the effects both of motor and of receptor fatigue. 
He investigated a comparatively simple instinctive action, the back wip- 
ing movements of the frog (Rana esculenta). He transected the frog's 
medulla oblongata in such a way that breathing remained undisturbed 
while spontaneous movements of the limbs were eliminated. The "wip- 
ing reflex" by which anurans remove disturbing stimuli from their dor¬ 
sal side could still be elicited, although it was slightly weakened imme- 
diately after the transection. Some "training" was necessary before the 
fixed motor pattern had regained its former response level. Using the 
frog thus prepared as object, Franzisket examined the question as to what 
extent the readiness to perform a certain motor pattern is specific to that 
pattern; in other words, to what extent can it be influenced by fatiguing 
similar and functionally analogous movements. Two different motor pat- 
terns were investigated: first, the common and well-known wiping 
reflex, in which the hind foot is brought onto the back at the rear end 
and moved forward so that the toes scrape along the entire back as well 
as over the head; second, the heel wiping, in which the heel of the foot 
is moved as far forward as it will go on the frog's back and is then scraped 
over it in a caudal direction. The fatiguing of one of these motor patterns 
proved to have no effect whatever on the readiness to perform the other, 
as is shown in Figure 10. 

The second important question put by Franzisket was "whether the 
amount of excitability is generally correlated to a type of movement, or 
whether it is a phenomenon inherent to the nervous pathway along 
which the response takes its course" [my translation]. The two wiping 
movements just mentioned offer a good opportunity to examine this 
problem because both of them can be elicited by stimulation at two phys- 
ically and neurologically distinct body areas. These areas are divided by 
an externally visible ridge and each is supplied with its own sensory 
nerves, the nervi cutanei dorsi mediales and the nervi cutanei dorsi 

Franzisket first released a series of toe-wiping movements through 
stimulation of the dorsal locus until, after a number of responses, the 
wiping movements became rarer (Figure 10). Then he transferred the 
point of stimulation to the lateral part of the frog's back and obtained a 
much smaller number of responses, yet slightly more than he would 
have got without changing the locality of stimulation. The reverse exper- 

6. Action-Specific Fatigue 


T oe wipe reflex 

Toe-wEpe reflex 

kbhh imt.... - i m r.rm 

Toe-wipe reflex 

■ ■! ¡ ■.■ . i ■ i i ¡ i ¡ i [ r ¡ n ¡ i t ]. r l ■ j i u 

■ m i ■■ i || i i i i i i i n i i || 

Heei-wipe reflex 

Figure 10. 200 scratch stimuli, applied to the same place on the back, elicit many 
toe-wiping responses among the first hundred and only a few such responses 
among the second hundred (two rows at top). When, however, the second 
hundred stimuli are applied to the caudal area at the tail end of the trunk for a 
release of the heel-wiping reflex, the large number of heel-wiping responses (bot- 
tom row) indicate that the preceding toe wiping movements have not influenced 
the amount of the capacity for excitation available for the heel wipe. (Franzisket, 
L.: "Untersuchungen zur Spezifitát und Kumulierung der Erregungsfáhigkeit 
und zur Wirkung einer Ermüdung in der Afferenz bei Wischbewegungen des 

iment, stimulating first laterally and afterwards dorsally, brought the 
same results. It is to be concluded that the motor pattern itself possesses 
a specific amount of excitability, as the change of receptor pathway can- 
not appreciably increase the number of movements released. Still, the 
slight increase of response caused by the change of locus seems to indi- 
cate that receptor fatigue could play its part, too. 

For a clearer demonstraron of receptor fatigue, it was necessary to ren- 
der the motor sector of the wiping movement "indefatigable"; Franzisket 
achieved this by allowing the frog to rest for ten days instead of two 
days—as had been done during the earlier experiments. Given this much 
rest, the motor sector for the pattern was not exhausted after 100 stimu- 
lations, the number which had become standard during the preceding 
experiments. But now fatigue on the receptor side became apparent. If 
stimulation was given only at the same locus, at the standardized interval 
of ten seconds, the response began to wane after 50-80 performances, at 
a time when the motor readiness was not yet diminished, as could readily 
be demonstrated by changing the stimulus locus, as is shown in Figure 
11 . 

When, instead of changing the locus of stimulation, Franzisket inter- 
posed a pause of 30 minutes, he obtained the same eífect as that achieved 
by changing the receptor pathway. Obviously, recovery from complete 
exhaustion takes a much shorter time on the receptor side of the wiping 
response than on its motor side. At the same time, the excitability of the 
motor pattern can be accumulated and preserved over a much longer 
period than that of the receptor mechanism. From his study of fatigue 
and recovery which, in the aíferent and the eíferent sectors of the system 
proceed with characteristically diíferent speeds, Franzisket has drawn 
the following conclusions: The quantity of excitability of the motor pat¬ 
tern is a function of central nervous structures (supposedly a trophic State 


I. Fixed Motor Pattern 


E I ■ : ■ j . i t :■ ■■ , rwn — 

Figure 11. Seven consecutive stimulus series, one following immediately after 
another, which were alternately applied to the back (Stimulus Area I) and to the 
flank (Stimulus Area II) for a release of the toe-wipe reflex. Because of a rest 
period of ten days prior to the initiation of this experiment, the appearance of 
numerous wipe responses, up to about the middle of the fourth stimulus series, 
is indicative of the accumulation of the capacity for excitation. (Franzisket, L.: 
"Untersuchungen zur Spezifitát und Kumulierung der Erregungsfáhigkeit und 
zur Wirkung einer Ermüdung in der Afferenz bei Wischbewegungen des 

in the connecting neurons coordinating the motor patterns), while 
fatigue is inherent to the afferent pathway (supposedly a function of syn- 
apses between the afferent path and special groups of connecting neu¬ 
rons which achieve the coordination of the wiping movement). An argu- 
ment for assuming that the excitability of the wiping movement is due 
to a trophic State of the connecting neurons coordinating it lies in the 
fact that the speed of its recovery after complete exhaustion is dependent 
on temperature (1953). The most important conclusión is that the concept 
of a specific quantity of excitability obviously corresponds to a physio- 
logical reality. 

The organism has at its disposal a limited amount of excitability spe¬ 
cific to a certain motor pattern. The quantity of this amount is as different 
in homologous patterns of different animáis as it is in different motor 
patterns of the same species. The metachronous coordination of segmen- 
tal muscles that eífects the swimming movements of a mackerel or of a 
a pelagic shark is as indefatigable as are the contractions of a bird's heart. 
The beating of the pectorals of a wrasse (Labrus) is practically inexhaus- 
tible as long as the fish is awake; the homologous movement made by a 
sea horse (Hippocampus) can be sustained, at the most, for a number of 
successive minutes only. The "supply" available for these movements is 
correlated exactly to the "demand" made on the organism, and this is 
expressed as the average amount of time it has to spend performing 
them. It is obvious why locomotion must be difficult to exhaust in so 
many creatures. In the pelagic fish already mentioned, locomotion is so 
constant that they can aíford to abolish the mechanisms which, in other 
animáis, serve for breathing: their unceasing forward movement causes 
a constant stream of water, sufficient for oxygenation, to pass through 
their gills. These fish suífocate if their forward locomotion is prevented. 

7. Threshold Lowering of Releasing Stimuli 


Another teleonomic reason for having an inexhaustible supply avail- 
able for locomotion is its importance for escape. For a majority of instinc- 
tive motor patterns, specific exhaustion sets in long before the organism, 
as a whole, is completely exhausted. Very few instinctive motor patterns 
can be repeated right up to the limits set by the basic functions such as 
breathing, heartbeat, or general muscular fatigue. Yet the readiness to 
flee, the overpowering urge to get away, is still at work even when the 
peripheral organs fail to obey. From the teleonomic viewpoint, the pri- 
macy of escape is obvious. 

As will be discussed in Section 12 of this chapter, locomotion is sub¬ 
ordínate to quite a number of superior "commanding instances," each of 
which puts it to the Service of another teleonomic function. A roebuck 
must run in order to escape from the wolf, in order to pursue a doe, in 
order to chase away a defeated rival, in order to seek a better pasture, 
and so on. A similar dependence on múltiple motivations is found in 
many instinctive movements. A mouse must gnaw in order to open hard 
seeds, to excávate a burrow in compacted soil, to free itself from a trap, 
et cetera. I have suggested calling these kinds of motor patterns "tool 
activities" (Werkzeugbezvegungen) because, like simple tools, they can 
serve more than one function. Perhaps the term "multipurpose move¬ 
ments" would be preferable. 

The fact that these activities are, for the most part, performed under 
the influence of a superior motivation does not by any means imply that 
they are lacking in a spontaneity proper to themselves. On the contrary, 
they possess a particularly strong endogenous production of specific 
excitability which is indeed very necessary because they must be availa- 
ble at any time and at short notice. More than other instinctive motor 
patterns, multipurpose movements show a quick lowering of thresholds 
whenever they are not used for a time. Also, they tend to be performed 
in vacuo more often than any others. The wolf running endlessly and 
aimlessly to and fro within its enclosure, the ostrich incessantly plucking 
nonexistent grass, the rodent gnawing all objects within reach, are famil¬ 
iar to every zoo visitor. 

7. Threshold Lowering of Releasing Stimuli 

While in Section 5 of this chapter I described the experiments made by 
Seitz, I have not as yet placed suíficient emphasis on one of their most 
important results: When the fighting movements of Seitz's Haplochromis 
had not been elicited during a period considerably exceeding the average 
intervals between experiments, the fish would not only be ready to per- 
form fighting movements with greater intensity and continué to do so a 
greater number of times, but it would also respond to a much weaker 
stimulation. Franzisket would probably have found the same thing if he 
had investigated a gradation of stimuli possessing a quantifiable releas- 


I. Fixed Motor Pattern 

ing effect. Working as he did with a single standard stimulus, he dem- 
onstrated only the increase in internal readiness and not the concomitant 
lowering of releasing thresholds. 

The recovery of internal readiness to perform a certain instinctive 
motor pattern after exhaustion does not, as in other cases of recovery 
after fatigue, stop at a certain predetermined level, but continúes to build 
up as long as the experimental animal is deprived of adequate stimula- 
tion under otherwise biologically desirable circumstances. Wallace Craig 
(1918) was the first to describe this phenomenon in the motor patterns 
of courtship of the blond ring dove (Streptopelia risoria). A male of this 
species, after having been isolated from conspecifics for some time, was 
ready to court a domestic pigeon, something he had refused to do earlier. 
After isolation had been continued even longer, the bird was ready to 
court a human hand; still later, he directed his courtship activities toward 
the rear córner of his box where the convergence of the three edges 
offered at least a point for fixation. In most cases of alleged "vacuum 
activities" a similar substitute can be found if one looks for it. Most inten- 
tional crossings of different species are achieved by breeders through the 
simple method of isolating the animáis long enough to cause the thresh¬ 
olds of their sexual responses to be lowered sufficiently to accept a non- 
conspecific partner. In dummy experiments Lissmann studied the pro- 
cess of threshold lowering of fight-eliciting stimulus configurations in 
the Siamese fighting fish (Betta splendens), with results analogous to those 
obtained by Craig in his investigations of courtship in doves. 

The thresholds of all stimulus configurations releasing the same set of 
motor patterns fluctuate parallel to each other, exactly as does the readi¬ 
ness of the several activities belonging to that set. In many cases the 
intensity of appetitive behavior directed at their performance also fluc- 
tuates in parallel. There is hardly any doubt that these three kinds of 
phenomena are consequences of an identical physiological process. In 
our institute's jargon it has become customary to describe the complex of 
these phenomena of accumulation as the "damming up" (Stau) of an 
instinctive action. This is, of course, a rather inexact conceptualization, 
because to what extent the observable phenomena are due to processes 
in the afferent, and to what extent in the motor section of the system, 
ought to be investigated in every individual case. The experimental 
results of the work done by Seitz, Franzisket, H. Prechtl, and others tell 
us that both may be the case. We also know that the endogenous pro- 
duction of excitability often takes place in the receptor itself, but we do 
not know to what extent this may be specific for an instinctive action. 

With regard to the accompanying subjective phenomena, it is trivial to 
State that the "damming up" of accumulated readiness first becomes 
noticeable on the receptor side. If, while on holiday in the south of 
Europe, we pass by a butcher's shop when we are hungry, we perceive 
the appetizing smell of raw beefsteak; but when we pass the same shop 

8. Effects Obscuring the Accumulation of Action-Specific Excitability 


after having dined well, we unambiguously smell carrion. A tramp 
steamer captain who was a friend of Oskar Heinroth once asked him to 
explain the following experience: Wherever he returned to Hamburg 
after a very long voyage, all the women in that city appeared to him to 
be very beautiful, but after he had been in port for some time, most of 
them seemed quite ugly. Goethe, in Faust, comments on the same phe- 
nomenon by saying: "Du siehst mit diesem Trank im Leibe bald Helenen 
in jedem Weibe." ("With this potion inside you, you will soon see Helen 
of Troy in every woman.") 

It seems to be an unanswerable question whether it is possible to 
remove or lower accumulated action-specific thresholds by repeated pre¬ 
sentaron of releasing stimuli without any consumption of motor readi- 
ness, since the latter is influenced by many kinds of stimulation. 

8. Effects Obscuring the Accumulation of Action-Specific 

If, in order to study the increase of appetitive behavior, motor readiness, 
and threshold lowering, one keeps an animal under conditions that pre- 
vent the arrival of specifically releasing stimuli, it is not so very easy to 
avoid two side effects obscuring the phenomena. 

The first of these is a true atrophy of the activity caused by its remain- 
ing unused for too long a time. After Heiligenberg kept a male Pelmato- 
chromis cribensis isolated for a very long time, the fish, on being presented 
with a conspecific rival, at first appeared to be extremely ready to fight, 
but contrary to all expectations and unlike Franzisket's frog, it was 
exhausted more quickly than it would have been after a shorter period 
of stimulus deprivation. This lack of stamina, however, disappeared after 
a short period of training. 

A more lasting atrophy of a motor pattern caused by long disuse is 
exploited by aviculturists to induce full-winged waterfowl to stay where 
one wants to keep them. If one clips the wings of such a bird before it 
has fledged, one renders it incapable of flight until the next wing molt. 
After such treatment and for the rest of its life, the bird will be less ready 
to fly than it would have been if allowed to fledge normally. By clipping 
the wings of the bird again before it can fly after the molt, the effect is 
even more pronounced and more lasting. A female greylag that had been 
kept clipped by her former owner until she had reached her third year 
always flew much less than any of the other healthy geese. She was never 
seen circling high in the air; she flew only in order to get from one place 
to another. With regard to muscular power and flying technique she was 
in excellent form and she demonstrated this by "whiífling" and also by 
coming down steeply and landing in difficult places. 

A second effect which tends to obscure the accumulation of action-spe- 


I. Fixed Motor Pattern 

cific excitability is the waning of general arousal. For a long time I doubted 
the validity of the concept of general arousal, but there can be no doubt 
that higher organisms are subject to threshold fluctuations which con¬ 
cern not one, but many or all of an animal's responses to external stim- 
ulation; a human can be sleepy when overtired, or generally aroused 
after having drunk too much coffee. When subjected to what are conven- 
tionally called constant and controlled laboratory conditions, an animal 
is usually deprived of a great part of the kind of stimulation that does 
not specifically elicit any particular response, but which still contributes 
to keeping the animal awake. At an early date I recognized the fact that 
the constancy of the environment in which most or all captive animáis 
must live is a dangerous pathogenic factor because it tends to slow down 
metabolism as well as all nervous processes. In my attempts to keep tame 
animáis in perfect condition, I long ago devised adequate methods to 
keep them sufficiently "amused"—which is to say, generally aroused. 

In her experiments on the marine Jewel Fish (Microspathodon chrysu- 
rus), O. A. E. Rasa (1971) made a special study of the phenomena of de- 
arousal occurring under constant conditions. She had to do this in order 
to demónstrate the accumulation of specific excitability that was being 
counteracted by the unspecific waning of general excitability. In the 
small tanks in which the fish were experimentally isolated, their general 
excitability diminished rapidly, that is to say, all thresholds, even those 
of mutually exclusive responses, rose parallel to one another. The result- 
ing States were, in their higher degrees, strongly reminiscent of sleep. All 
locomotion became rare and was slowed down; the coloration of the fish 
paled in so exact a correlation to the other phenomena of de-arousal that 
it was proved that color could be used as a measure of the State of excit¬ 
ability prevailing at any moment. Unspecific changes in the environ¬ 
ment, such as a new piece of coral in the tank, stronger lighting, water 
currents caused by a pump, or even a Christmas tree ornament hung 
before the front pane of the tank, all had the eífect of temporarily elim- 
inating the de-arousal, that is, of lowering all threshold valúes. 

The assumption of a variable State concerning general excitability is 
not new to physiology. Stellar (1954) supposed that general arousal 
might be controlled by a certain area within the hypothalamus; R. Jung 
demonstrated by electrical stimulation that, in the cat, the formatio retic- 
ularis can cause a rise of general excitability. And although it was Rasa's 
work that caused me to realize the existence of this important phenom- 
enon, I have since been told that H. W. Magoun was the first to express 
these ideas in 1950. 

Rasa's demonstration of the influence which unspecific stimulation 
exerts on the general State of excitability is of great theoretical impor- 
tance because it draws attention to a similarity in the functions of stimuli 
which generally "charge up" the readiness to act and those of stimuli 
which specifically release activity. Later, in Two/IV/4, I shall discuss 

9. Vacuum Activity 


these problems further and I will try to show why the concepts of excit- 
ability and of excitation are difficult to define. 

9. Vacuum Activity 

If one keeps an experimental animal under tolerably good conditions 
that do not cause any great loss of general excitability but which, at the 
same time, are so contrived as to exelude the stimulation that, under nor¬ 
mal conditions, would be adequate to release a certain instinctive behav- 
ior pattern, the phenomena already described appear, that is, increased 
appetitive behavior, the lowering of stimulus thresholds, and an increase 
of readiness to perform the pattern in question. With some species and 
with some motor patterns this process can reach an extreme, and the pat¬ 
tern can "go off" without any noticeable external stimulation. It is obvi- 
ously impossible to say "without any stimulation" because, as in the case 
of Craig's ring dove which directed its courtship behavior at the meeting 
point of three straight lines, a substitute object is easy to find, even if it 
is an extremely weak one. Also, as has already been mentioned and as 
will be fully discussed in Two/III/3, 4, it is not possible to draw a sharp 
line between those stimulus configurations that "charge up" readiness 
for a certain pattern and those that set it going. So our definition of what 
we cali a "vacuum activity" remains, necessarily, inexact. 

Vacuum activities are most striking when they involve motor patterns 
which, under normal circumstances, are directed at an object. I shall 
never forget the behavior of a starling I kept when I was a schoolboy. 
One day while sitting on the head of a bronze statue in our dining room, 
the bird kept scanning the white ceiling in an excited manner. From time 
to time it took off, flew up to the ceiling, snapped at something, carne 
back to its perch, performed the movements of beating an insect against 
the perch, then swallowed and appeared satisfied for the moment. I had 
to climb up on pieces of furniture and even then I had to fetch a ladder 
before I had really convinced myself that there were no flying inseets in 
that room. At the age of 17, Bernhard Hellmann and I fully realized the 
importance of this observation. It may be of historical interest that our 
Germán term, auf Leerlauf, did not mean 'in vacuo' at all, but was a term 
taken directly from motoreyeling parlance, from the expression of letting 
the motor run 'in neutral'. In fact, this is a much better metaphor for 
what actually happens. Another striking example of a vacuum activity is 
the weaverbird's (Quelea) nest building. If the males of this species are 
kept in a cage devoid of anything that can be used as nest-building 
material, they will still perform the very complicated motor patterns of 
tying blades of grass to the twigs they perch on. 

In these cases of a motor pattern obviously directed at a certain object 
but performed in spite of its absence, the naive observer definitely 


I. Fixed Motor Pattern 

receives the impression that the animal is hallucinating the missing object. 
This interpretation is not as farfetched as it may seem. What happens in 
the central nervous system of an animal performing a vacuum activity 
may differ from the teleonomic performance of the same pattern only 
with regard to a few, obviously dispensable reafferences. How then 
should the poor creature know the difference between dream and 

There are cases in which an animal cannot be deprived of releasing 
situations because, if all others are lacking, parts of the animal's own 
body can be used as substitute objects. A female canary (Serinus canarius) 
which I isolated for quite different purposes performed the actions of 
collecting nesting material, carrying the material to the nest site, and of 
building there by grabbing her own breast feathers in her beak, carrying 
them to a córner of the cage, and performing the "tremble-shove" move- 
ments that will be described later. Young rats, which Eibl-Eibesfeldt iso¬ 
lated in order to study their nest-building activities, used their own tails 
as nesting material: they searched for material to carry, "found" their 
own tails, carried them to the nest location and carefully deposited them 

From my own observations I know of three cases in which aggressive 
behavior patterns were directed at an animal's own body, using it as a 
substitute object. Bankiva cocks (Gallus bankiva), reared in total isolation 
by Kruijt (1971), persisted in attempting to hit their own tails with their 
spurs. They kept squinting backwards at their long tail feathers; sud- 
denly they would jump round by 180° to strike the air where their tails 
had just been. Mallard drakes (Anas platyrhynchus) treated in exactly the 
same way by Schutz (1965) also tried to attack their own tails and they 
persisted in this stereotyped activity for a long time even after having 
been liberated on the Ess-See. A test question for visiting ethologists 
then was how the crazy circlings and bitings backward of these drakes 
should be explained. I once possessed a Moorish idol (Zanclus canescens) 
which used to attack its own tail whenever it made a sharp turn and its 
tail carne into its field of visión. Then it would circle for a time, in much 
the same way that the mallards did. Normal fighting in this species 
involves a circling of the adversaries around each other, and this differs 
from the behavior just described only in its larger radius. 

The frequency with which a motor pattern occurs as a vacuum activity 
is clearly correlated to the frequency with which it is normally per¬ 
formed. This has to do with the question of "demand for" and "supply 
of" a certain instinctive movement and will be discussed later. Gallina- 
ceous birds of various families have a motor pattern consisting of scrap- 
ing with their closed beaks to uncover food particles, and captive birds 
do this almost incessantly on the floors of their cages. Bankivas and other 
members of the crested family scratch with equal persistency with their 
toes. Rodents, mice for instance, are forced by their endogenous impulse 

10. Appetitive Behavior 


production to spend much of their time gnawing, quite independently 
of whether gnawing is necessary or not. 

The behavior patterns most frequently appearing in the form of vac- 
uum activities are, for obvious reasons, those of locomotion. Interestingly 
enough, in birds as well as in mammals, the readiness for locomotion is 
closely bound to the readiness for escape. From the point of view of 
internal readiness, flight appears as the highest intensity of locomo¬ 
tion—which again is teleonomically understandable. As every eques- 
trian knows, an "accumulation" of locomotion through keeping a horse 
stabled for some time increases not only its readiness to bolt, that is, to 
run fast, but also its readiness to buck, jump sideways, stop suddenly 
with head lowered, and every other stratagem that has been pro- 
grammed in the phylogeny of horses for throwing oíf large predatory 
mammals. The exuberance which calves, lambs, kids, or young rabbits 
show after being liberated from cióse confinement is closely akin to the 
behavior of the horse that has been stabled for too long: they not only 
indulge in some fast running, but they also perform zig-zag and side¬ 
ways jumps and other motor patterns that serve in escaping from pred- 
ators. Birds of very different taxonomic orders show the same gradations 
of motor patterns beginning, at a low intensity, with mere locomotion, 
and culminating, at the highest degree of intensity, in motor patterns 
whose teleonomic function indubitably is the avoidance of flying pred- 
ators, in particular of raptors stooping toward the bird from above. Birds 
as different as ravens, swallows, and geese fly oíf faster and faster when 
liberated after a period of captivity and, having gained sufficient altitude, 
perform the same and probably homologous motor patterns of turning 
on their backs, presenting the supporting areas of their wings upwards 
to the air current, and dropping earthwards with an acceleration exactly 
double that which could be achieved through the use of gravity alone. 
This motor pattern is probably common to all carínate birds and is reg- 
ularly seen during States of "exuberance." 

These dramatic eífects of a short period of "damming up" locomotion 
should remind us once again of the fact that the motor patterns involved, 
although often "commanded" by other motivations, possess a very high 
production of their own action-specific excitability. 

10. Appetitive Behavior 

At a time when I myself had fully grasped the implications of what had 
been observed with regard to threshold lowering, to endogenous accu¬ 
mulation of excitability, and to vacuum activities, I had not yet quite 
realized the importance of another essential consequence of stimulus 
deprivation. It was Wallace Craig who first drew attention to the fact that 
a motor pattern which the animal had not been able to discharge for 


I. Fixed Motor Pattern 

some time not only causes threshold lowering and an increase of excit- 
ability, it actually creates an excitation of its own which activates the 
whole organism and causes the animal to search for releasing stimuli. In 
the simplest case, this kind of disquietude causes the organism to move 
aimlessly and at random, but even this undirected restlessness serves to 
increase the probability of the animal's meeting the stimulus configura¬ 
ron that can release the undischarged motor pattern. In its most compli- 
cated forms it involves learning and purposive behavior. In fact, we do 
not know any instance of learning by reward—conditioning by rein- 
forcement—which does not take place within the context of searching 
or striving behavior that is always determined by a phylogenetically 
evolved program. In his classic paper, "Appetites and Aversions as Con- 
stituents of Instinct," Wallace Craig (1918) called this ubiquitous phe- 
nomenon appetitive behavior. 

As an opposite to appetitive behavior or appetence, Craig conceptual- 
ized aversión. While appetence continúes until a certain stimulus is 
reached, aversión continúes until a certain stimulus has been got rid of, 
a stimulus that Craig, following Thorndike, called the "original 
annoyer." Craig's concept of aversión is not quite as clearly defined as 
his concept of appetence. As Meyer-Holzapfel (1956) has shown, the 
avoidance behavior embraced by Craig's term must be separated into two 
concepts, a) that of true aversión, and b) that of an appetitive behavior 
aimed at quiescence. Because the differences between these two processes 
lie mainly in their effects on conditioning, they will be discussed in more 
detail in Three/IV/3, 4. Here it suffices to say that true aversions play an 
important part in all conditioning by punishment. 

At present we know only a very few innate motor patterns which do 
not give rise to appetitive behavior when the animal is deprived of the 
releasing stimulus situation for an appreciable period. One such is the 
guttural gobble noise made by the male turkey: even in a stage of strong 
threshold lowering, the bird does not show any searching for releasing 
stimuli. A motor pattern which does not lead to threshold lowering 
when "dammed up" is not known to us; no behavior with a constant 
threshold has ever been observed in animáis. 

11. Threshold Lowering and Appetitive Behavior in 

The consequence of withholding the adequate object of an instinctive 
motor pattern for an unusually long time can have a teleonomic effect: 
intensified searching must improve the probability of finding adequate 
stimuli and the lowering of thresholds must diminish selectivity so as to 
make it possible for the organism to accept, in a pinch, an object that is 
not quite up to standard. There is a proverb in Germán: In der Not frifit 

11. Threshold Lowering and Appetitive Behavior in Avoidance 


der Teufel Fliegen ("Other food lacking, the devil eats flies.") At least this 
is obvious for all action patterns that commence with organisms turning 
toward the object of the instinctive activity. 

With behavior patterns of avoidance, the phenomena described in the 
last section appear, at first sight, to be dysteleonomic. Particularly the 
incessant fluctuating of thresholds seems to serve no sensible end. Work- 
ing in the field with tame full-winged wild geese, one is made most pain- 
fully aware that the thresholds of the stimuli that elicit wild panic and 
escape in these birds are subject to the wildest fluctuations. At one 
moment a slight rustling of dry grass or a small downy feather drifting 
along on a breeze may cause a great flock of geese to rise thunderously 
into the air, while, at another time, a dog running through brushwood 
or the supernormal dummy of a flying predator, a hang glider rushing 
by at low altitude, may cause an alarm of much lesser intensity, such as 
a stretching-up of necks and a following of the object with the eyes. If 
one considers the enormous waste of energy these birds can ill aíford to 
expend, which is incurred by the repeated and unnecessary taking wing 
of a great flock and, on the other hand, the equally enormous danger of 
reacting in too leisurely a way to an approaching danger, one cannot 
help wondering why it is apparently impossible for the mechanisms of 
evolution to construct, in the greylag goose (Anser anser), a releasing 
mechanism responding with a constant threshold to the stimuli emanat- 
ing from the only flying predator endangering their species, the white- 
tailed eagle (Haliaetus albicilla). 

Intraspeciíic aggression is generally counted among avoidance 
responses because it serves, under normal conditions, the dispersal, that 
is, the regular distribution of individuáis over the habitat available to the 
species. The defenders of stimulus-response psychology tend to deny 
spontaneity of behavior in general, and strong ideological reasons 
demand that this denial should be emphasized with regard to aggressive 
behavior. In particular, the existence of appetitive behavior aimed at the 
discharge of íighting has been grimly negated by authors who should 
know better. A man "looking for trouble" is, perhaps, not too familiar a 
sight in civilized circles, but a greylag gander or a zanclus (Zanclus canes- 
cens) in this mood is a spectacle that can be seen practically every day. 

Many years ago Lissmann demonstrated the lowering of the threshold 
releasing íighting behavior in the Siamese íighting íish (Betta splendens). 
A really active search for a situation for íighting first caught my own 
attention through the íish already mentioned, Zanclus canescens. In a 
large tank a dominant íish had intimidated a weaker íish so completely 
that the weaker íish was constantly hiding behind a rock wall at a place 
where it was not only invisible to the dominant íish but also unreachable 
except by means of very steep detour that, moreover, led very cióse to 
the water's surface where it crossed the upper edge of the rocks. Corning 
cióse to the water's surface constitutes a strong deterrent to most íish. 


I. Fixed Motor Pattern 

The dominant zanclus moved around quietly in his home range near the 
tank's front pane. From time to time his intention to attack became visi¬ 
ble through the laying back of the dorsal and anal fins in preparation for 
a strike and, at almost the same moment, he would rush toward the 
invisible victim to harass it still further. I have seen this very same seek- 
ing a fight with an invisible opponent innumerable times in the same 
species and under much more favorable conditions in my very large tank 
at Altenberg. 

On these observations of the appetite for agonistic behavior in Zanclus , 
Rasa (1971) based her study of the same phenomenon in the marine 
Jewel Fish (Microspathodon chrysurus). She showed that the appetite for 
agonistic encounters could be used just as well as the appetite for food as 
a reward situation for teaching a fish to swim through a simple maze. 
The fish were made to learn a simple L-shaped maze at the end of which 
the reward consisted of being able to see a conspecific antagonist 
through the diaphanous walls within a plástic chamber. The fellow 
member of the species put into the chamber was chosen so as to be 
slightly smaller than the experimental animal in order to invite attack. 
To get measurements of motivation it proved necessary to limit the size 
of the experimental chamber so as to make staying in it sufficiently dis- 
agreeable for the attacker; otherwise the fish would remain in it indefi- 
nitely, glaring at its unattainable opponent. 

The amount of time the fish spent in the chamber was clearly corre- 
lated to the intervals during which it had previously been prevented 
from entering the chamber. In the evaluation of these experiments, the 
effects of general de-arousal (mentioned in Two/II/8, and to be further 
discussed in Two/III/4) were either prevented or taken into considera¬ 
ron and compensated for. 

As Hogan and Adler (1963) also demonstrated in the fighting fish 
(Betta splendens), the presentation of a fighting opponent has exactly the 
same effect, as a reinforcement for the learning of a maze, as has the pre¬ 
sentation of food. Even without any systematic experimentaron, anyone 
who is sufficiently familiar with the natural behavior of fish or of birds 
knows how strong the appetite for agonistic behavior can be in very 
many species. In the swordtail (Xiphophorus helleri) Franck and Wilhelmi 
(1973) studied the effects of "damming up" agonistic behavior patterns 
and found all the typical phenomena of increasing readiness and 
decreasing thresholds, as well as clearly defined appetitive behavior. 
Polemics should be kept out of a textbook, but it must be mentioned here 
that the editor of a well-known Germán journal specializing in popular 
Science thought it fitting to head the publication of Franck's and Wil- 
helmi's results with the title, Spontaneitat der Aggression — Nein! (Spon- 
taneity of Aggression—No!), thus implying the exact opposite to what 
was being demonstrated by their paper. The ideological bias against all 
spontaneity that emanates from stimulus-response psychology is sur- 
prisingly strong in itself, and this bias rises to something resembling 

12. Driving and Being Driven 


religious fervor the moment the spontaneity of aggressivity enters into 
the consideration. This is exasperating to anyone who, through lifelong 
observations devoid of any underlying hypothesis, has become familiar 
with higher animáis such as fish, birds, and mammals and has encoun- 
tered, innumerable times, unambiguous evidence of the spontaneity of 
agonistic behavior. Why indeed should it be less spontaneous than any 
other kind of behavior? It would be extremely surprising if it were. 

12. Driving and Being Driven 

In principie, all instinctive motor patterns are capable of furnishing their 
own autonomous contribution to the múltiple motivations which keep 
an animal going. Equally, all processes based on endogenous generation 
of impulses, including complex instinctive action patterns, are suscepti¬ 
ble to additional stimulation coming from other sources of motivation. 
This is true at many different levels; even the excitation-producing loci 
of the heart are spontaneously driving and, at the same time, susceptible 
to being driven. The pacemaker producing the excitation that causes the 
atrium and the ventricle of the heart to contract possesses its own rhythm 
in discharging endogenous impulses. Nevertheless, the heart still beats 
at a considerably higher rate because impulses coming from another 
superordinated pacemaker, situated on the sinus, always arrive a fraction 
of a second before those coming from the atrioventricular pacemaker are 
discharged. The rhythm of the superordinated, faster pacemaker is, in its 
turn, subject to accelerating and inhibiting influences exerted by the ner- 
vus accelerans and the nervus depressor cordis. 

In this respect the instinctive activities of animáis "behave" very much 
like the pacemakers of the heart: although possessing their own endog¬ 
enous production of excitation, each of the motor patterns hitherto inves- 
tigated in this respect acts as a drive, in the accepted sense of the word. 
At the same time, each of the motor patterns is subject to "being driven" 
by other factors. Some motor patterns, however, are more resistant to 
being driven than others. Highly differentiated specific motor patterns, 
those of copulation for instance, which are performed rarely and only in 
one equally specific stimulus situation and only in the Service of one 
function, are largely independent of other motivations and are never 
performed except for their own sake, that is to say, never in the Service 
of, or activated by, any other system. Unless one conceives of hormonal 
States as stimulus situations, they are very resistent to "being driven." As 
will be discussed later in Three/IV/5, these one-purpose activities are 
also very difficult to condition. The other end of a long scale of increas- 
ingly "driveable" motor patterns is represented by the multipurpose 
activities which, as already mentioned, are programmed in such a way as 
to serve very different motivations. 

In principie, all instinctive motor patterns are exactly what A. F. J. Por- 


I. Fixed Motor Pattern 

tielje, in his book. Dieren zien en leer en kennen (1938), has called "action- 
and-reaction-in-one." This term, although rather lengthy even without 
the Dutch double vowels, fits the facts as no other. The quantity of 
impulses endogenously produced during any time span (as already men- 
tioned in Two/I/6), is adapted to the average demand which the species 
in question has for that particular motor pattern. This is not only true on 
the low level of fin movements in fish; a small bird such as a tit (Parus), 
which uses its wings every few moments as long as it is awake, com- 
mands an almost unlimited supply of flight movements that, therefore, 
appear as "voluntary" to the human observer. A greylag goose, in con- 
trast, takes wing comparatively rarely and can easily get into a situation 
of "wanting" to fly, but being unable to do so because its internal readi- 
ness is insufficient at that moment. If, when returning from an extensive 
flight, a goose has the bad luck to land within a narrow enclosure from 
which it can escape only by taking wing again, it shows at once that it 
possesses full insight into the situation. The goose does not try to find an 
exit through which it might walk out; it incontinently proceeds to per- 
form preflight movements. But it cannot take oíf simply because it is not 
able to raise its specific excitation for flying above the necessary thresh- 
old. The observer is strongly reminded of a man very near the threshold 
of sneezing but who fails to bring oíf the relieving explosión; the man 
tries to reach the threshold through self-stimulation—for instance, by 
looking at a strong light. The goose, in the situation described, also 
resorts to self-stimulation, to calling, and to bill-shaking; but often a long 
time elapses before it can finally take off. The bird is unable to make a 
decisión (the Germán word Entschluf3 is even more descriptive etymolog- 
ically), and the movements of flying obviously are not "voluntary" in 
the sense discussed in Three/V/2. 

Paul Leyhausen has studied the endogenous impulse production of the 
several motor patterns pertaining to prey-catching in catlike carnivores 
(Felidae). He put a hungry cat in a room that was completely devoid of 
any cover but filled with an abundance of live mice, thus maximally 
facilitating prey-catching. In every one of these experiments, the cat first 
caught, killed, and ate half a dozen of the mice, then killed a few more 
without eating them, then proceeded "playfully" to catch some more 
without executing the killing bite. After reaching this stage, the experi¬ 
mental cat would sit quietly in the attitude of lying in ambush, with its 
head lowered and intently watching some of the mice running about on 
the opposite side of the room while others were actually crawling, unno- 
ticed, over its paws. The number of performances after which each of 
these motor patterns proved to be exhausted corresponds exactly to the 
relative frequency with which each of them is "in demand." A cat often 
has to spend a long time sitting in ambush before being oífered an 
opportunity to stalk a mouse, and usually it must do a lot of stalking 
before it ever gets into a position from which the long leap at the prey 
has any likelihood of success, and the leap can miss so that several leaps 

12. Driving and Being Driven 


must be programmed in relation to each killing bite. Even this can fail 
to accomplish its end, so the cat must have more killing bites at its dis- 
posal than the number of mice it can eat. 

In the complex hierarchical systems serving various animáis during 
the acquisition of food, it is by no means the process of actually eating 
that invariably constitutes the most important consummatory act. In 
birds of prey, as falconers know, the main motivation of food acquisition 
is the catching and not the eating of the prey. There seems to be a gap in 
the innate program between the sequence of catching and that of eating: 
Having caught its first prey, a naive peregrine falcon (Falco peregrinus) 
appears emotionally exhausted and rather at a loss over what to do next. 
Then it begins plucking the prey in a tentative, haphazard way. Actual 
pecking, that is, the beginning of eating, seems to set in only under the 
influence of the optical stimulus of seeing blood. In dogs, hunting and 
killing are similarly independent of the motivation of eating: As every- 
one knows, it is impossible to wean a dog from its passion for hunting 
through abundant feeding. The consummatory act toward which the 
hunting appetite is striving is the killing, particularly the shaking of the 
prey, and not the devouring of it. 

Neweklowsky (1972) studied the manner in which the prying move- 
ment of the starling (Sturnus vulgaris) is dependent on internal and exter- 
nal influences. The starling thrusts its bilí into clefts or into soft material 
and then opens it with considerable muscular forcé, thereby enlarging 
or making a cleft into which it is able to peer. In birds possessing this 
prying movement, and in adaptation to this motor pattern, the cleavage 
between the mandibles is positioned so as to be in an exact straight line 
with the pupil of the eye, so that the line of visión leads unimpeded into 
the fissure opened by the prying movement, however narrow this may 
be. This organization has evolved at least three times independently: in 
some starlings (Sturnidae), in some icterids (orioles, blackbirds, meadow 
larks) and, curiously enough, in the penduline tit (Remiz pendulinus) 
which uses the movement only for the building of its elaborately woven 

The prying of the starling is strongly stimulated by the sight of a nar¬ 
row cleft and by the touch of soft edges offering only limited resistance 
to being widened. Tame starlings seem to enjoy it enormously if one per- 
mits them to thrust their bilí between two fingers that offer only a little 
resistance to being pried apart. Also, the starling quickly learns to rec- 
ognize objects soft enough to be torn by prying, such as a newspaper 
lying on a soft support. The frequency of prying is increased only to a 
small degree by hunger; the increase is just at the borderline of being 
significant. On the other hand, exploratory behavior (Three/V/2) con¬ 
tributes appreciably to the motivation of prying: Any new object offered 
to a starling elicits long bouts of prying. Otherwise, the external stimulus 
situation has little influence on the number of prying movements per- 
formed within the time unit. In a cage completely devoid of fissures or 


I. Fixed Motor Pattern 

clefts to pry into, and with abundant food openly offered, the frequency 
of prying is not significantly lower than when the starling has to acquire 
all of its food by prying into cups closed by rubber diaphragms that have 
slits in them and are not too easy to pry open. In a species that, like the 
European starling, acquires most of its food by performing one single 
motor pattern, the autonomous motivation of the latter is obviously suf- 
ficient to ensure adequate feeding and needs no appreciable additional 
drive from the side of tissue need. 

The cactus finch (Cactospiza palliáa) of the Galápagos Islands acquires 
a considerable part of its food by means of an extremely complicated 
behavior pattern that ineludes the use of "tools": the bird grabs a sharp 
cactus spine of suitable length in its bilí and thrusts this into holes and 
crevices to poke inseets out of them. Like the prying movement of the 
starling, this motor pattern is performed most persistently even when no 
prey can be acquired by it and when sufficient food is offered in open 
dishes. A strong appetitive behavior is demonstrably aimed at the inter- 
mediary goal of the spine, the particular releasing properties of which 
Eibl-Eibesfeldt has investigated. A tame cactus finch will greedily grab at 
a nice spine and immediately begin to poke it into clefts even when some 
favorite food, such as a wax moth larva, is offered simultaneously. Given 
the choice between a good spine and a fat larva, even a rather hungry 
cactus finch will unhesitatingly grab the former. Eibl-Eibesfeldt has even 
observed one of his finches put a larva into a hole, search for a spine and 
then poke the larva out again—in a way similar to a sportsman stocking 
his "shoot" with pheasants. 

13. Neurophysiology of Spontaneity 

In 1935, when I gave that lecture at the Harnack House in Berlin, I was 
already aware of all the salient faets concerning the spontaneity of 
instinctive motor patterns although not of their quantitative verification, 
which has been mentioned here. I discussed all the effeets of threshold 
lowering, of vacuum activities and of appetitive behavior, and I even 
quoted from the letter written to me by Wallace Craig in which he said 
that it was "obviously nonsense to speak of a re-action [ sic ] to a stimulus 
not yet received." All this knowledge notwithstanding, I still stubbornly 
upheld the theory that instinctive movements were based on chain 
reflexes, although not on simple linear ones but on complicated polysy- 
naptic networks of reflex processes. I assumed that the phenomenon of 
spontaneity—which I had conscientiously described—could be 
explained by an auxilliary hypothesis. 

In those amazingly unenlightened assertions, to recapitúlate briefly 
what has been written in the Introductory History, I was motivated by 
two prejudices. One was the belief that the obviously mechanical perfor¬ 
mance of an instinctive movement and the equally reflex-like reaction to 

13. Neurophysiology of Spontaneity 


a very specific stimulation were arguments in favor of the chain reflex 
theory. The second was an almost subconscious antagonism to the theo- 
ries of vitalists such as McDougall (1923) and Bierens de Haan. I stupidly 
felt that relinquishing the chain reflex theory of instinctive actions 
meant a concession to purposivistic, that is, vitalistic psychology, and 
such a concession was something I was unwilling to make. 

Nowadays, as I have already explained, it seems obvious that I should 
have asked the following question: Do we know any physiological pro- 
cesses that, though certainly as mechanical as instinctive movements 
and, like these, certainly not requiring any vitalistic explanation, but 
which, nevertheless, are not reflexes, which in fact are demonstrably 
independent of external stimulation and tend to recur rhythmically at 
regular intervals? The obvious answer would have been that processes 
of this kind have been known for a very long time in the excitation-pro- 
ducing loci of the heart. 

Production of impulses within the heart was by no means the only 
process of endogenous generation of stimuli known at that time. In par¬ 
ticular, the results obtained by Erich von Holst on the endogenous pro¬ 
duction and coordination of motor impulses within the ventral cord of 
the earthworm, as well as in the spinal cords of fishes, were quite unam- 
biguously analogous to those obtained by Craig, Seitz, and myself on 
intact organisms. In fact, all of the phenomena discussed in the previous 
sections—with the exception of appetitive behavior—have been dem- 
onstrated by von Holst on a lower level of neural organization. It is sur- 
prising that his work is hardly ever mentioned in some of the very recent 
books on ethology. 

Paradoxically, the very first investigation by von Holst that was des- 
tined to explode the explanatory monopolism of the reflex concept was 
originally aimed at analyzing an alleged "chain reflex." At the time, the 
following experiment was regarded as the paradigm and final proof for 
the concatenation of reflex processes and was even quoted by Jakob von 
Uexküll despite his fundamental antipathy to mechanistic explanation. 
An earthworm was cut in two and the pieces were subsequently tied or 
sewn together with thread. The worm then continued to creep along 
with well-coordinated movements, the waves of successive contractions 
running over the segments of its body and passing over the cut without 
any apparent hindrance. This was regarded as proof that the passive 
stretching which each segment underwent because of the contraction of 
the segment preceeding it caused its own contraction by way of a reflex. 
It was assumed that the swimming movements of a fish, for instance 
those of an eel, were coordinated in the same way; the contraction or 
relaxation of one muscle segment was believed to influence the muscular 
tensión of the following one by means of proprioceptors which released 
reflexes. The way in which von Holst proceeded shows, most surpris- 
ingly, that he must have known that all these explanations were wrong 
even before he started experimenting. 


I. Fixed Motor Pattern 

His first experiments were based on the following considerations. If 
the conduction of excitation really consists of a chain of successively 
released reflexes, it must fulfill a number of conditions. Excitation must 
always reach one segment after the other just as, in a row of standing 
bottles, one must fall after the other when the first has been knocked 
over. But this is not always the case; under the infiuence of ether all the 
segments of a worm's body extend simultaneously. Furthermore, excita¬ 
tion should prove incapable of passing over a stretch of the ventral cord 
from which all peripheral nerves have been cut, since the path of the 
reflexes has been interrupted. What actually happens is that the excita¬ 
tion passes over such isolated stretches of the ventral cord with even 
greater rapidity than it does over the intact ones. 

After these preliminary experiments, von Holst completely isolated a 
worm's ventral cord from its body, removed the supra-esophageal gan- 
glion, suspended this preparation in Ringer's solution, and connected 
every single one of the ventral ganglia to a highly sensitive voltmeter. 
These instruments showed spikes of a tensión, but not randomly—which 
would have been important enough in itself—but in a regular sequence; 
beginning at the front end, the wave of excitation passed over the entire 
length of the preparation at approximately the same speed and with the 
same rhythm in which the wave of contraction passes over the intact 
worm's body while it is creeping. Von Holst went on to prove that the 
impulses made visible by means of the instuments were indeed identical 
to those causing segmental muscle contractions: he made similar prepa- 
rations in which a few segments of the ventral cord were not isolated 
and connected to voltmeters but were left in connection with the muscle 
segments. These then contracted in perfect coordination with the seg¬ 
ments whose ganglia were connected to voltmeters. 

Similar experiments on spinal fishes had similar results. Using a tench 
(Tinca tinca) with all posterior sensitive nerves cut away from the spinal 
cord so that reflexes could no longer be released from the skin, the fish 
could still react to optical and static stimulation and could swim with 
normal, well-coordinated movements. 

An eel treated the same way also retains a normal coordination of the 
swimming movements. If the middle part of its long body is immobilized 
by cutting not only the sensitive posterior roots but the anterior motor 
roots as well, the wave of the swimming undulation, beginning at the 
fish's front end, disappears into the paralyzed part as a train would into 
a tunnel, only to reappear at the non-paralyzed tail end of the fish with 
the phase relationships nearly unchanged—not quite unchanged 
because, for reasons unknown, the processes of the centrally coordinated 
wave of excitation pass through the immobilized part of the body at a 
rate slightly fáster than normal. The same effect is attained if a part of 
the eel's body is immobilized, not by cutting nerves, but by putting it 
into the strait jacket of a narrow tube. 

As all these experiments show, none of the motor patterns investigated 

13. Neurophysiology of Spontaneity 


is based on a chain of reflexes but on processes of spontaneous genera- 
tion and of coordination of impulses; both take place within the central 
nervous system itself, through processes independent of extero- or pro- 
prioceptor stimulation and of segmental reflex ares. 

With regard to the normal undulating swimming movements of fish, 
simple observation could have pointed out the fact that, when a fish 
starts swimming, its movements involve, from the very beginning, all 
the segments of its body simultaneously, and that stopping is accom- 
plished in the same way when the fish arrests all its swimming move¬ 
ments. According to the reflex theory, undulation ought to begin at the 
front end when the fish starts swimming, and gradually pass along and 
out at the tail end when the fish stops. 

After having demonstrated to his own satisfaction the spontaneity of 
endogenous stimulus generation within the central nervous system, von 
Holst went on to investígate the factors responsible for the coordination 
of movements. Since movement coordination is independent of afferent 
nerves, the problem for investigation was obvious but there were some 
difficulties in achieving its solution. One of these lay in the fact that the 
motor pattern presented itself as a fully integrated whole the very 
moment it was initiated. As soon as an organism changes from a State of 
quiescence into one of motion, coordination is perfect and, as von Holst 

. . . it is as impossible to draw any conclusión concerning its génesis as it is 
impossible for the experimental embryologist to obtain any information 
concerning the factors causing and directing its ontogenetic development 
by the study of the mature organism. This difficulty forces us to look for 
objeets in which the finished, absolute coordination of the several motor 
organs does not represent the only relation that is possible between them, 
but rather represents an extreme limiting case—the other extreme being 
represented by complete lack of any mutual influence between motor pro¬ 
cesses. Perhaps such objeets could show us all transitions, all grades of 
mutual influence, ranging from very slight interactions to the strongest 
obligatory linkages [my translation]. 

Von Holst found the desired object for his investigations in the swim¬ 
ming movements of those fishes which, for propulsión, use rhythmical 
skulling movements of the fins rather than the more common method of 
undulating the whole body so that the tail fin serves as the main pro- 
pellent. Wrasse and some other percoid fishes use tail fin swimming only 
during emergencies demanding máximum speed; otherwise they propel 
themselves by means of a rhythmical paddling of their soft fins while 
keeping the body stiff. Analyses of films taken of intact fish show that 
the fins often beat synchronously and in absolute coordination for a long 
time, and then this coordinated connection is broken quite abruptly and 
the fins beat completely independently of one another, each fin having 
its own rhythm; that is to say, there is no coordination at all. The most 
frequent and, at the same time, the most interesting relationship between 


I. Fixed Motor Pattern 

the rhythmical beating of two fins is the following: Each fin tends to 
maintain its own frequency but each exerts a certain quantitative influ- 
ence on the the beating of the other, both with regard to its frequency 
and to its amplitude. The physiological mechanism eífecting this phe- 
nomenon has been termed "relative coordination" by von Holst. 

Von Holst developed a simple method for recording the fin move- 
ments under the influence of relative coordination. After transecting a 
fish's medulla oblongata and arranging for artificial respiration, he con- 
nected each fin to a lever writing on a rotating drum. The lever, con- 
structed from extremely thin slivers of straw, was connected to only a 
few rays of a fin, the others having been cut away in order to eliminate 
the eífects of water resistance. "By this procedure," von Holst says, "an 
unpredictable creature is suddenly transformed into an apparatus of high 
precisión which performs movements of complete regularity as long as 
external and internal conditions are kept constant, but which reacts to 
each external influence with a definite change in its activity." 

To evalúate the curve obtained by this method, von Holst used a pro¬ 
cedure based on the principie of Fourier's sequences, a discussion of 
which here would lead too far afield. Every biologist and particularly 
every student of behavior interested in the physiology of the central ner- 
vous system is urgently advised to read the papers by Erich von Holst 
(1973) listed in the Índex of this book; they are available in paperback 

For a majority of cases in which the rhythms of two fins stand in a 
relationship of relative coordination, these rhythms affect each other 
mutually, although one of them often has a stronger influence than the 
other. In such a case, one can speak of a dominant and of a dominated 
rhythm. In an extreme case, one rhythm can exert a very one-sided dom- 
ination over the other without, in turn, being appreciably influenced by 
it. In labrids and probably in many other perch-like fish, the rhythm of 
the pectoral fins dominates all the others. The two pectorals stand in a 
relationship of absolute coordination with each other. This is so regular 
in a majority of fish that the one known exception strikes every observer 
familiar with fish behavior as humorous: the pectorals of the Moorish 
idol (Zanclus canescens), by moving without any coordination at all, con- 
vey the compelling impression of an inebriated fish. In most perch-like 
fish, the pectorals have two motor patterns and both of them are executed 
in absolute coordination on both sides. In one motor pattern, both fins 
move in a skulling movement, pressed broadside against the water and 
pushing the fish forward while, during the return stroke, they move 
edge on. In the other pattern, the fins also move in absolute coordination 
but alternately and with an undulation taking place among the single fin 
rays, producing a downward directed stream of water. The first coordi¬ 
nation serves to propel the fish forward, the second allows it to hover in 
one place. It is only this second coordination of the pectorals that influ- 
ences the beat of the other two fins. 

13. Neurophysiology of Spontaneity 






Figure 12. Labrus: movement of one pectoral fin (upper trace), the second pectoral 
fin (middle trace) and the dorsal fin (lower trace). In a , the two pectoral fin 
rhythms are inhibited by pressure on the sides of the body; in b , only one pectoral 
rhythm is inhibited; in c, the two pectoral fins oscillate in alternation. (von Holst, 
E.: The Behavioural Physiology of Animáis and Man , Volume One.) 

The influence which the pectoral fins exert on the subordínate 
rhythms of the caudal, of the soft dorsal, and of the anal fins can be dem- 
onstrated with particular clarity through the trick of excluding the pec¬ 
toral rhythms for a time. This can be done quite simply by compressing 
the anterior part of the fish's body. The effects of the re-awakening pec¬ 
toral rhythm can be observed as it gradually takes charge again. As Fig¬ 
ure 12 shows, the dominated rhythms beat in a regular sinusoidal curve 
as long as the dominating rhythm is excluded. This demonstrates what 
form the dependent rhythms would assume if left to themselves and also 
what changes are caused by the dominant rhythm when it assumes com- 
mand again. 

The influence of an absolutely dominant rhythm may here serve to 
explain this effect because it is the simplest form of relationship between 
two or more endogenous rhythms. As already stated, the influences 
which two rhythms exert upon each other are, as a rule, mutual. Each 
rhythm has the tendency to maintain a constant frequency of the fin 
under its command—it shows a tendency to persevere (Beharrungsten- 
denz), as von Holst called it. The effect of the dominant rhythm on the 
dependent rhythm can be described as an attraction exerted by the peaks 
of the dominant one on those of the dependent one: If the culmination 
of the dominant fin movement is reached just before that of the depen¬ 
dent fin, the latter is speeded up; if the dominant fin lags behind, the 
beat of the dependent fin is slowed down. This effect gives the impres- 
sion that the dominant rhythm is "trying" to forcé its own frequency on 
that of the dominated rhythm. The forcé of this influence is dependent 
on the span between the culminations; the nearer they are to one 
another, the stronger is the attraction—much like the effect a magnet has 
on iron filings, and for this reason von Holst coined the term "magnet 


I. Fixed Motor Pattern 

b ^\rv/v 



Figure 13. Scheme for explanation of Figure 12. See text for details. (von Holst, 
E.: The Behavioural Physiology of Animáis and Man, Volume One.) 

effect" (Magneteffekt). Necessarily, with two rhythms beating at different 
frequencies, the phase relationships change periodically. When the 
phase of the dependent rhythm precedes that of the dominant rhythm, 
the latter has the effect of slowing down the dependent rhythm while, 
on the other hand, the frequency of the dependent rhythm will be accel- 
erated if it is just a little behind that of the dominant one. The strength 
of this influence—and this is essential—is changing periodically with 
the periodic changes of phase distances between the two rhythms; it is 
always increasing when the distance between culminations decreases 
and vice versa. 

Obviously, the quantity of this influence must be the greater the more 
often the culminations of the two rhythms coincide or are near each 
other, and this happens most often when the frequencies stand in a rela- 
tionship that can be expressed in whole numbers, for instance, if every 
second beat of a dependent rhythm comes near to a beat of the dominant 
rhythm. At a certain, very small distance, the dependent rhythm often 
catches up with the dominant one by making a little jump—which is 
exactly what is so strongly reminiscent of the effect of a magnet. 

Besides, and together with, the magnet effect, there is another form of 
mutual interaction between two endogenous rhythms—one that is more 
easy to understand and to explain. This is simple superposition. Again, 
and for simplicity's sake, the rare case of absolute one-sided dominance 
of one rhythm over the other is chosen as an example. As Figure 13 
shows, the amplitude of the dependent rhythm increases whenever it 
beats in the same direction as does the dominant one, and exactly to that 
extent corresponding to the peak of the excursión made by the dominant 
rhythm at that very moment. Therefore, the fin activated by the depen¬ 
dent rhythm describes a curve corresponding to a superposition of the 
two curves of both rhythms. 

Relationships between two rhythms based exclusively on superposi¬ 
tion are just as rare as those based on the magnet effect alone; in the great 
majority of instances both effects are at work simultaneously. Also, a 
purely one-sided dominance is just as rare with regard to superposition 
as it is with regard to magnet effect; in fact, it is only the rhythm of the 

13. Neurophysiology of Spontaneity 


pectorals which, in the fish species investigated by von Holst, proved to 
be absolutely dominant. 

An interesting detail is the following. As already mentioned, the pec¬ 
torals of a labrid and acanthurid fish beat alternately as long as the fish 
is standing still, but synchronously when it is swimming forward. It is 
only in the former case that the rhythm of the pectorals exerts a domi¬ 
nant influence on the other fins. When the pectorals beat synchronously, 
the rhythm of the dependent fins becomes regular, freed from superpo- 
sition and the magnet effect, not because the influence emanating from 
the pectoral rhythm ceases, but because the effects of the two pectorals 
now beating in opposite directions neutralize each other. To the best of 
my knowledge, von Holst never discussed the question of which influ- 
ences are at work between the two pectoral rhythms: They always beat 
synchronously (except in Zanclus ); it is only the sign of their excursión 
peaks that changes. 

There are all kinds of possible intermediates and transitions between 
relative and absolute coordination, but the phenomena of magnet effect 
and of superposition are quite sufficient to explain them all. It is easy to 
understand why, for aquatic animáis like fish, relative coordination is 
sufficient to produce tolerably regular motor patterns of locomotion; 
even in an untrained swimmer, relative coordination between arm and 
leg movements is a common occurrence. For locomotion on dry land, less 
changeable phase relationships between limb movements are desirable 
for the simple reason that there, in addition to propulsión, the functions 
of supporting the body and of maintaining equilibrium become impor- 
tant. Nevertheless, relative coordination occurs in terrestrial animáis 
more often than one would suppose. Dogs, for instance, besides their 
absolutely coordinated paces of galloping, trotting, and walking, often 
show transitions into relative coordination when they change from one 
pace to another. When changing from walking to trotting, some dogs 
show the normal trotting coordination, in which contra-lateral legs are 
moved simultaneously, but they regularly fall into the coordination of 
the "ambling" trot, in which ipsilateral legs move together, when they 
change from galloping to trotting. They do not continué ambling for 
very long, however; they soon return to normal trotting by way of an 
interposed period of relative coordination. 

These and other phenomena are in agreement, down to the smallest 
detail, with our assumption that identical physiological processes 
underly relative coordination in aquatic and absolute coordination in ter¬ 
restrial organisms. The magnet effect appears to have a still wider distri- 
bution. It has been demonstrated in the aguish movements of patients 
afflicted with Parkinson's disease and it is almost banal to State that it 
also prevails in voluntary movements; everyone knows how difficult it 
is to play 4/4 with one hand and 3/4 with the other on a piano. 

In Two/I/2, 3, when speaking of intensity differences of instinctive 
motor patterns, it was mentioned that two different central nervous pro- 


I. Fixed Motor Pattern 

cesses cooperate in centrally coordinated movements. By a method and 
a reasoning whose discussion would lead too far afield here, von Holst 
has shown that there is a duality of functions, one sort of cell producing 
rhythmic impulses and several kinds of motor cells, each of which 
responds to these impulses with a slightly different threshold. The cells 
producing endogenous impulses determine the rhythm and its fre- 
quency, while the amplitude of the movement is dependent on the num- 
ber of motor cells which are activated, "recruited" by these impulses. An 
explanation of the phenomena here described would indeed be most dif- 
ficult if one should assume that the motor elements which send their 
impulses directly to the muscles were identical to those responsible for 
the endogenous generation of impulses as well as for their coordination. 
On the other hand, says von Holst, "the facts are easily explained if one 
assumes that two kinds of elements are contained within the spinal cord, 
one of which generates endogenous rhythmical impulses to which the 
other kind, the motor cells proper, respond by being alternately excited 
and inhibited and by sending corresponding motor impulses to the mus¬ 
cles." When the relative coordination of two rhythms is at work, the 
amplitude of the fin movements activated by the dominant rhythm fluc- 
tuates parallel with the influence it has on the dependent rhythm, in 
other words, parallel with the measure of the latter's periodicity. This 
influence becomes noticeable even when, for instance, the pectoral fin, 
after having been inhibited, begins to beat with a very small amplitude, 
although with the characteristic frequency which it will retain later. The 
gradual increase of amplitude is caused, as already explained, by the 
"recruitment" of motor cells which, owing to their different thresholds, 
respond in increasing numbers to the increasing intensity of endogenous 
impulses. Because of this the interaction of several rhythms can produce 
a constant phase relationship, in other words, a "gestalt" or a "melody" 
of the overall movement that remains recognizable even at very different 
intensities, as has been mentioned in Two/1/2, 3. All these complex 
and harmonious movements are produced entirely without the help of 
proprioceptors and even without the help of afferent processes in 

Endogenous production of impulses has since been demonstrated to 
exist in very many different organisms, in very different parts of the cen¬ 
tral nervous system and even in non-nervous tissues, such as in tissue 
cultures taken from the heart of chicken embryos, in the body wall of 
coelenterates (Batham and Pantin 1950 a,b) and in the muscles of poly- 
chaete worms (Wells 1950). It exists in the central nervous system of the 
crayfish (Astacus) (Prosser 1961), in that of mantids, grasshoppers and 
cockroaches (Roeder et al. 1960), in isolated pieces of the brain cortex of 
cats (Kristiansen and Courtois 1949), in cell cultures made from the gan- 
glion cells of Aplysia, and in all sensory epithelia that have been inves- 
tigated hitherto. 

14. Analogies of Function in Neural Elements and Integrated Systems 


14. Analogies of Function in Neural Elements and 
Integrated Systems 

The central nervous system plays what may be called the "dirty trick" of 
accomplishing, on very different levels of integration, analogous func- 
tions in so perfectly similar a manner that even a highly sophisticated 
scientist can be misled into thinking them to be physiologically identical. 
The classic example of such a mistake was made by Helmholtz in assum- 
ing that the abstracting functions of perception mentioned in One/II/3 
were based on rational, although unconscious deductions. The greatest 
caution is necessary whenever one attempts to find a connection between 
properties inherent in neural elements and those properties found in 
complex systems into which these elements are built. 

This necessary reserve notwithstanding, it remains an important fact 
that even the smallest neural elements, down to isolated nerve fibers, 
show a very similar spontaneity as do whole animáis. The oíd reflex the- 
ory propounded, and some modern stimulus-response psychologists 
assume that nerve cells and whole organisms, like well-bred children of 
Victorian times, "don't speak unless spoken to," in other words, that they 
remain passive and silent until a stimulus of sufficient strength impinges 
on them. As Bullock remarked, "somehow it did not impress behaviorists 
that the fly on the table sometimes takes off without any apparent stim¬ 
ulus" (1977). The neural element is regarded as a rather mechanical con- 
traption that does not possess any more "behavior" than an electric 

The stimulus-response doctrine is all the more surprising since it is by 
no means certain that a neural element possessing the properties postu- 
lated by the doctrine exists at all. In every single instance in which a 
neural element has actually been investigated it has, as Kenneth Roeder 
pointed out, "been shown to have an elabórate intrinsic behavior deter- 
mined by membrane properties, cell geometry, ion concentraron, neu- 
rohumoral processes, and the spatial and temporal configuraron of extra- 
cellular impacts" (1955). 

Even in isolated neurites it can be demonstrated that the same element 
can be alternately reactive and spontaneously active. Figure 14 shows a 
comparison between the fluctuations of excitability in a spontaneously 
active and in a reactive element. On the ordinate axis is indicated the 
excitability prevailing at the moment. Its valué can only be measured by 
the strength of the stimulus necessary to cause the element to discharge 
a nervous impulse. The upper horizontal line represents the threshold 
which must be reached for this to happen, the lower horizontal indicates 
the valué of the rest excitability at the moment, for instance, in the neu- 
rite of a mammalian sciatic nerve suspended in Ringer's solution. The 
drawn-out curve shows the fluctuations of excitability which follow on 
the impact of the stimulus S. After a quick rise of excitation, the neurite 


I. Fixed Motor Pattern 


Figure 14. (Loiver graph) A comparison of the excitability changes in stable (stim- 
ulus-response) and unstable (spontaneously active) neural elements. Solid curve 
represents excitability sequence in elements such as mammalian A fibers follow- 
ing stimulus Broken curve represents predicted excitability sequence in ele¬ 
ments with a lower threshold and, henee, repetitively active following stimulus 
S v S 2 is the stimulus interpolated at some point in the spontaneous excitability 
change and causes premature discharge. (Upper graph) Nerve impulses that would 
arise from the excitability changes shown below. Time units would have different 
valúes for different types of excitable tissue. In the case of mammalia A fibers, the 
units would be milliseconds. (Roeder, K.: "Spontaneous Activity and Behavior ") 

"fires off" an impulse and then its excitability sinks to the valué of zero; 
in other words, no stimulus whatsoever can make the neurite discharge 
an impulse again for a certain duration. This time span is called the 
absolute refractory period. Subsequently, excitability rises again, in an 
undampened curve, overshooting the mark of rest excitability by far, 
regaining its valué only after some time and by way of a dampened 

A small change in the valué of rest excitability, easily effected by 
diminishing the calcium contení of the Ringer's solution, has the con- 
sequence that the excitability which follows on the refractory period not 
only exceeds the valué of rest excitability but actually reaches that of the 
threshold, thus causing another discharge and, therewith, a sequence of 
spontaneous, rhythmically discharged impulses. 

As will be discussed later in Two/III/2, it has been demonstrated that 
sensory cells—formerly regarded as paradigms of elements passively 
waiting for stimulation—are characteristically spontaneously active 
units. Pumphrey has convincingly indicated the teleonomy of this fact 
(1950); the spontaneously active element does not have a threshold in 
the conventional sense of the word. A stimulus may arrive at any time 
and be of any strength; even if it is extremely small it will cause a mod- 
ulation of the frequencies sent out by the receptor element. Today we 
know that the signáis of all sensory cells hitherto investigated do not 
consist of a single impulse but always of modulations on the frequeney 

14. Analogies of Function in Neural Elements and Integrated Systems 


of the impulses they produce spontaneously. Even for the neuron, and 
even for a neurite separated from its neuron, Portielje's oíd assertion 
made for intact animáis remains equally true: Its function is "action-and- 
reaction-in-one" (1938). 

In higher multicellular animáis all spontaneous activities, down to 
those of the smallest elements, are influenced by those of superordinated 
loci which, in turn, are effected by exterior factors. It is only in organisms 
devoid of a centralized nervous system that all spontaneous activities are 
directly influenced from without. It is worth some contemplation that 
mobile unicellular animáis behave so very much as metazoa do. 

If stimulus-response processes really played the role they are sup- 
posed to play according to the oíd reflex theory, one should expect that 
an organism which happens to be in an environment offering too little 
stimulation would get stuck at dead center, as Roeder says, and stay 
quiescent until a change in the environment could offer further stimu¬ 
lation that would set it going again. Very few animáis actually wait for 
stimuli or remain quiescent until stimuli arrive, although some highly 
specialized and cryptically colored predators seem to lapse into a natural 
kind of "akinesis" while lying in ambush waiting for prey. The larva of 
the ant lion (Myrmeleon) and some bottom-dwelling íish offer examples 
of this, but even they will move spontaneously if they must wait too 
long. For the vast majority of free-moving organisms, William 
McDougall's oíd saying is perfectly true: "The healthy animal is up and 
doing" (1923). 

On the basis of their investigations of the behavior of sea anemones, 
Batham and Pantin conclude that "whatever may be the origin of these 
activities, it appears to lie in the animal itself and not to be caused by 
external stimulation" (1950 a,b). There are any number of motor 
responses which we are wont to describe as "reflexes" because the 
sequence of their muscle contractions is as short as their coordination is 
simple and, also, because their reactivity is much more evident than the 
participation of spontaneous elements. In my opinión, it is permissible 
to speculate that some of them, such as blinking, swallowing, stretching, 
sneezing, and others, may be constructed in a way similar to fixed motor 
patterns. Many of them, such as blinking, swallowing, and others, can 
appear as displacement activities in situations of conflict. Stretching, 
sneezing, and yawning, like some typical instinctive activities, are built 
on the "orgasm principie," which is to say that action-specific excitation 
rises to a climax, then the consummatory act is discharged, after which 
the act cannot be repeated. As F. Beach demonstrated in the chimpanzee, 
a retrograde inhibition is exerted by proprioceptors reporting the con- 
summation. Having stretched or yawned, one cannot do either again 
immediately. Sneezing even has its own appetitive behavior, as is dem¬ 
onstrated by people buying snuff. 

This is not true of other reflexes such as the optomotoric, muscle 
stretch, vestibular-ocular, and many other responses serving to supply 


I. Fixed Motor Pattern 

instant information concerning body position and equilibrium. Many of 
these must have a constant or well regulated gain (input: output funo 
tion); also, their functioning should not leave any lasting engram 
because what they report must be open to being countermanded within 
the next fraction of a second. Many of these, such as the allegedly mon- 
osynaptic tendón reflex, are based on complicated regulatory systems. 

Apparently, responses which are not "action-and-reaction-in-one" are 
rather rare. Still, the element which lacks spontaneity and "does not 
speak unless spoken to " does exist. One such element, which W. Heili- 
genberg told me about, is the constricting muscle around the outflow 
opening of sponges. Also, the giant fibers which in many phyla cause 
extremely fast startle responses may, for long perods, show no vacuum 
activity, although the threshold is often quite labile (personal commu- 
nication from T. Bullock, 1981). 

What holds true, on the lowest level of integration, for the functional 
properties of neural elements and also, on the highest level, for the 
instinctive behavior of higher animáis, is equally valid for the interme¬ 
díate level of integration represented by the endogenously generated 
and centrally coordinated movements investigated by Erich von Holst. 
All attempts to explain fixed motor patterns and their properties on the 
basis of reflexes (however complicated and polysynaptic the hypotheti- 
cally chosen patterns may be) are so obviously constrained and far- 
fetched that their ideological motivation becomes apparent at once. If, on 
the other hand, one assumes that endogenous impulse production and 
central coordination form the basis of most animal activities, one finds 
the most natural explanations for just those phenomena which present 
the greatest difficulties to the chain reflex theory; threshold lowering and 
vacuum activities cease to be paradoxes and become effects which would 
have to be postulated if they had not already been observed. As W. 
Schleidt has pointed out in his paper, "How 'Fixed' is the Fixed Action 
Pattern?" (1974), some of the changes of form which an instinctive motor 
pattern undergoes through fluctuations of intensity find an uncon- 
strained explanation on the basis of von Flolst's assumption of the dual- 
ity of stimulus-production and motor function and on the basis of his 
theory of recruitment of motoric elements: what remain constant are the 
relationships of phase and amplitude which constitute the harmony of 
the pattern recognizable even at the lowest intensities. 

15. Chapter Summary 

1. There are unchangeable, easily recognizable sequences of move¬ 
ments which, as properties of species, genera, families, and greater 
taxonomic units, are as reliable as criteria of evolutionary relation¬ 
ships as are any morphological characteristics or combinations of 
such. To these fixed motor patterns the concept of homology is equally 

15. Chapter Summary 


applicable. The distribution of fixed motor patterns within the taxo- 
nomic system, as well as the degrees of their similarities and dissim- 
ilarities within different taxonomic groups proves, beyond the neces- 
sity of any further confirmation, that these motor processes, in every 
detail of their forms, are anchored in the genes in exactly the same 
way as bodily properties are. 

These genetically coordinated movements often, but not always, 
form a functional unit together with the releasing mechanisms that, 
without previous experience, respond selectively to stimulus situa- 
tions in which the motor patterns are apt to fulfill their teleonomic 
functions. This type of functional unit has been called arteigene Trieb- 
handlung —an instinctive action, or literally, a "species-characteristic 
drive action"— by Oskar Heinroth. Fixed motor patterns and innate 
releasing mechanisms are, however, entirely different physiologic 
mechanisms which can be incorporated into complex behavior Sys¬ 
tems independently of each other and in varying combinations. 

2. Far from obeying an "all-or-nothing law", fixed motor patterns can 
appear in forms of all the possible transitions from a very slight 
indication, called an "intention movement" by Heinroth, to the full 
teleonomic performance that occurs at the high intensity of specific 
excitation. What remains constant throughout all fluctuations of 
intensity are the phase relationships as well as the proportions of 
their amplitudes. This makes the "melody" of the movements rec- 
ognizable throughout all of their intensities. 

3. In some instances, the same quality of specific excitation elicits dif¬ 
ferent motor patterns, each of which is correlated to a particular 
degree of excitation. 

4. It is assumed that these motor patterns, correlated to stages of inten¬ 
sity, are indeed activated by the same kind of qualitatively uniform 
excitation. This assumption is based on the following facts: a) all of 
these movements are elicited by the same configurations of external 
stimuli, in other words, by one IRM; b) they can also be elicited 
through electrical stimulation at the same loci of the hypothalamus, 
as von Holst has demonstrated with the chicken. Their scale of 
intensities, correlated to electric stimulations of different strengths, 
corresponds exactly with the scale of intensities observed in the 
intact organism; c) the thresholds of all the motor patterns of such a 
"set" fluctuate strictly parallel to each other; the exhaustion of one of 
them raises the thresholds of all the others; d) the transition from 
one such motor pattern to the pattern of next higher or next lower 
intensity is achieved without any time lag, while the change from 
one independent system of motor patterns to another independent 
system is subject to an "inertia" which causes a measurable pause 
between the activities. 

5. The intensity with which a fixed motor pattern is performed 
depends on two factors: a) on the internal readiness of the organism. 


I. Fixed Motor Pattern 

and b) on the effectiveness of the external stimulation. The obser¬ 
varon of a motor pattern being released by a certain stimulus situa- 
tion furnishes us, therefore, with an equation containing two 
unknowns. Alfred Seitz added a second equation by offering an 
animal, after each dummy experiment, a stimulus configuration 
known to exert the maximal releasing effect. In this way he ascer- 
tained how much specific readiness was "still left" after the first pre¬ 
sentaron of the dummy. This he called the method of dual quanti- 
fication. The laws which determine the decrease of internal readiness 
caused by each performance of a fixed motor pattern, and those other 
laws which determine the releasing effect of a stimulus configura¬ 
tion, could only be found by investigating both simultaneously. 

6. Each performance of a certain motor pattern decreases the organism's 
readiness to perform it again without, at the same time, fatiguing the 
animal as a whole, that is, without affecting its readiness to perform 
other motor patterns. One of the few exceptions is represented by 
the motor patterns of locomotion that pertain to escape; these can be 
released until the entire organism, particularly with regard to its res¬ 
piraron and circulation, is exhausted. 

7. The action-specific fatigue of one motor pattern, or of one "set" of 
motor patterns, is different from fatigue in general with regard to 
one important point: After general fatigue, the restitution of normal 
readiness to act reaches a definite level and stops there, while the 
experimentally enforced quiescence of a fixed motor pattern causes 
an almost unlimited lowering of the threshold valúes of releasing 
stimuli; this makes possible the acceptance as stimuli of inadequate 
substitute objects. 

8. If an experimental animal is kept for a long period under constant 
and strictly controlled conditions that prevent the arrival of releas¬ 
ing stimuli, the above-mentioned phenomena of increasing readi¬ 
ness for a certain motor pattern can be obscured by two effects. One 
of these is an atrophy which can affect an unused motor pattern in 
much the same way it affects an unused muscle; the second is a grad¬ 
ual waning of "general arousal," that is to say, a general raising of 
all thresholds, a condition that is closely akin to that of sleep. The 
study of general arousal and, more particularly, of its opposite, gen¬ 
eral "de-arousal" caused by a deficiency in unspecific stimulation, 
has contributed considerably to our understanding of the relation- 
ship between stimuli which "charge up" the readiness and those 
which directly release a motor pattern. 

9. If the experimental animal is kept under conditions that exelude 
directly releasing stimulus configurations but still supply the organ¬ 
ism with sufficient unspecific stimulation to prevent de-arousal, the 
threshold lowering for certain motor patterns can reach the extreme 
valué of "zero," that is to say, the motor patterns can be discharged 

15. Chapter Summary 


in vacuo without discernible releasing stimuli. This is particularly 
striking for object-directed motor patterns. 

10. A long-enduring State of quiescence of a fixed motor pattern not only 
leads to a lowering of the thresholds of releasing stimuli but also 
induces a State of general restlessness in the organism as a whole. In 
its simplest form, this phenomenon, by causing the animal to move 
about randomly, increases its chances for encountering a releasing 
stimulus situation, while in its most complex differentiation, the 
same phenomenon constitutes the motivation for purposive search. 
This type of behavior has been termed appetitive behavior by Wallace 
Craig (1918). All conditioning by reinforcement occurs ivithin a context of 
appetitive behavior , at least in non-human species. 

11. Threshold lowering and appetitive behavior also occur in motor pat¬ 
terns which, because of their teleonomic function, must be consid- 
ered as constituting avoidance behavior. In some instances such as 
intra-specific aggression and in some patterns of escape, this may 
lead to effects which are clearly dysteleonomic, particularly if, as in 
some types of aggressive behavior, intraspecific selection plays a part 
that is very dangerous to the survival of the species. In escape behav¬ 
ior, too, the phenomena of spontaneity often appear to be distinctly 
dysteleonomic. It would seem that evolution has been unable to con- 
struct an IRM possessing a constant threshold. This is all the more 
surprising as a number of other cases are known in which thresholds 
can be remarkably constant. Otherwise, constancy of perception is 
largely dependent on a highly complicated feedback mechanism 
sensing and regulating blood sugar, body temperature and many 
other factors. The fact that the motor patterns of fighting are subject 
not only to threshold lowering but also give rise to virulent appeti¬ 
tive behavior has met violent, ideologically motivated denial. 

12. Every motor pattern that is striven for by its own appetitive behavior 
constitutes an autonomous source of motivation for animal and 
human behavior. On the other hand, hardly any fixed motor patterns 
are known which cannot be "driven," that is, activated or facilitated 
by motivations coming from outside the realm of their own autono¬ 
mous spontaneity. This rule holds equally true for the lowest levels 
of motor patterns such as the rhythmic contractions of the vertébrate 
heart as well as for motor patterns demonstrably caused by endoge- 
nous generation and central coordination of impulses, and also for 
the most complex systems of instinctive behavior. A. F. J. Portielje's 
rather long term, "action-and-reaction-in-one," is perfectly fitting for 
all of these cases. It is only the quantitative measures of driving and 
of being driven that vary in the different cases. The amount of a 
motor pattern's susceptibility to "being driven," in other words, its 
"availability," is dependent upon the quantity of its endogenous 
generation of "readiness," of excitability. 


I. Fixed Motor Pattern 

13. All the facts stated and described in the preceding sections find an 
unconstrained explanation through the research results of Erich von 
Holst who demonstrated that, in the ventral ganglion chain of the 
earthworm (Lumbricus) and in the spinal cord of many fishes, the pro- 
cesses of endogenous generation of impulses and their coordination 
occur within the central nervous system itself without afferent stim- 
ulation taking any part in shaping the coordinated "melody of 
impulses." This explodes the chain-reflex theory of instinctive 

14. An important way in which two primarily independent processes of 
endogenous production of impulses can become coordinated has 
been called the "magnet effect" by Erich von Holst. Each rhythm 
influences the other in the sense that the peaks of the movements 
produced attract one another. This effect is the stronger the nearer 
the peaks are to each other. In consequence of this, the rhythms tend 
to forcé upon each other those frequencies which stand in a tempo 
relationship that is "harmonious," in other words, that can be 
expressed in simple whole numbers. If the frequencies of two 
rhythms are suíficiently cióse to a harmonious relationship, the mag¬ 
net effect forces them into keeping step permanently. Thus "relative 
coordination," found more often in aquatic animáis, is turned into 
"absolute coordination" which prevails in organisms moving along 
on dry ground. 

15. Notwithstanding the great caution that is indicated when applying 
conclusions drawn from the properties of the elements to properties 
of the system into which they are integrated, it is still permissible to 
point out that even the smallest units of nervous tissue, including 
even neurites (Roeder 1955), are far from conforming to the idea of 
the purely reactive elementary process postulated by the reflex the¬ 
ory. Slightly more integrated functions of central nervous systems, 
such as centrally coordinated movements (von Holst), show an even 
more impressive similarity to the behavior of intact animáis. All the 
phenomena of spontaneity which cannot be accounted for by the 
reflex theory become not only explicable but must be postulated if 
one assumes that the spontaneous generation and central coordina¬ 
tion of impulses, as demonstrated by Roeder and von Holst, form the 
basis of the fixed motor patterns observed in intact animáis. 

16. The essence of everything that has been said in this chapter can be 
condensed into Erich von Holst's analogy: "The central nervous sys¬ 
tem is not like a lazy donkey that must be whipped or, to make the 
comparison closer, that must bite its own tail every time before it is 
able to move a step; the central nervous system more closely resem¬ 
bles a spirited horse as much in need of the bridle as of the spurs." 

Chapter II 

Afferent Processes 

1. The Innate Releasing Mechanism (IRM) 

As has been emphasized repeatedly, the functional unit of the instinctive 
action now called the species-characteristic drive action consists of two 
fundamentally disparate physiological processes. The active, sponta- 
neous part was the first to get our attention; therefore, as I have tried to 
reconstruct the génesis of our own ideas in this book, the fixed motor 
patterns have been discussed before the afferent processes. As long as the 
whole of the motor activity was regarded as a chain of reflexes, the first 
link of this chain did not seem to be different from the subsequent ones 
ñor did it seem to demand any special attention. With increased insight 
into the physiological nature of spontaneous motor patterns, however, it 
became clear that it must indeed be a physiological apparatus of a very 
different kind, an apparatus which selectively "recognizes" the biologi- 
cally correct situation and, thereupon, removes the inhibition which oth- 
erwise blocks the performance of the fixed motor pattern. 

In the last sections of the preceding chapter I was forced, for reasons 
already explained, to anticipate some of the facts concerning the mech- 
anisms which enable the animal to "recognize," without any previous 
experience, a biologically relevant situation—in other words, to respond 
to it selectively by a teleonomically "correct" and equally unlearned 
action pattern. At first I called this selective afferent mechanism the 
angeborene Schema —the innate scheme—because the organism seemed to 
have some sort of simplified, sketchy information about what the biolog¬ 
ically relevant situation was like. Later, Tinbergen and I relinquished 
this term because its connotations suggested that something like an out- 
line or an image of the whole situation or object was innate. In discussing 


II. Afferent Processes 

the methods of dual quantification, I was forced to anticípate the impor- 
tant fact that it is by no means an image of the whole object or situation 
which is innately "known" to the animal, but a number of indepen- 
dently effective, very simple stimulus configurations whose releasing 
functions, obeying the law of heterogeneous summation, add up to a 
qualitatively unitary effect. For this reason Tinbergen and I (1938) aban- 
doned the term Schema and decided to cali the neural organization here 
under discussion the innate releasing mechansm (IRM)—-in Germán, ange- 
borener Auslósemechanismus (AAM). 

This concept is defined exclusively by its function. It is obvious that in 
organisms differing with regard to the complexity of their nervous orga¬ 
nization as well as to the levels of integration attained by their cognitive 
faculties and their behavior, very different demands are put upon the 
selectivity of their responses to external stimulus configurations. It is 
equally obvious that very different physiological mechanisms have 
evolved to cope with these demands. 

Even with the lowest organisms, the question arises concerning 
whence the animal obtains the information telling it what to do under 
which circumstances. In Two/VI, I shall describe in detail the mecha¬ 
nisms evolved to receive and exploit instant information. The IRM is to 
be regarded as one of these, yet the problem of its selectivity must be 
discussed here. How can we explain that, for instance, an amoeba does 
not ingest all corpuscles of suitable size but refuses—though with some 
pardonable errors—to eat undigestible material? We know that a flagel- 
late, obeying a kinesis (Two/IV/5), accelerates its locomotion when tra- 
versing unfavorable conditions and slows down on entering more favor¬ 
able ones and, by these simple reactions, achieves the goal of spending 
most of its time within the latter. But how does the flagellate know 
which environmental conditions are favorable and which are not? The 
system of actions, the repertoire of behavior patterns at the disposal of 
these protozoans comprises but a few possibilities for motor activities, 
among which the avoidance of danger and the approach to food are pre- 
dominant. Therefore, no excessive demands are made on the selectivity 
of the releasing mechanisms. Still, if an amoeba responds to quite a num¬ 
ber of quite different stimuli in a manner which—at least statistically 
considered—is sufficiently teleonomic to assure the survival of the spe- 
cies, this fact needs an explanation, all the more so since the motor 
response to all these stimuli, the amoeboid movement, remains virtually 
the same throughout, changing only in intensity and in symbols. All an 
amoeba can do is to creep, with varying intensities, toward or away from 
a source of stimulation, the extremes being represented by creeping "all 
over" an object, and thus engulfing it, or, in the opposite case, by thick- 
ening the ectoplasm of its entire surface, in other words, by capsulating. 
The identity of the motor mechanisms makes it all the more remarkable 
that a multitude of different stimuli evoke either approach or avoidance 

1. The Innate Releasing Mechanism (IRM) 


in the teleonomically correct situation. The mechanism underlying this 
selectivity is little known; Chemical stimuli are predominantly 
responded to, but thermical and tactile ones are, too. 

A little bit more is known about the releasing mechanisms by which 
ciliates, particularly paramecium and related forms, are guided. The most 
important criterion by which the environment is "judged" to be favora¬ 
ble or not, is its pH valué. The mechanisms of phobic and of topic 
responses, to be discussed in Chapter VI, have the effect of keeping the 
animáis in water with optimal acidity. The acid found much more fre- 
quently than any other under natural conditions is C0 2 . It is regularly 
found in the vicinity of decomposing organic material as a product of 
bacterial activity. This regularity is sufficiently reliable so that the ciliates 
can "afford" to be programmed to respond to certain concentrations of 
acid as a sign of the presence of bacteria on which they can feed. Under 
normal conditions all other acids are so rare that the survival of no spe- 
cies of ciliates is threatened by their reacting dysteleonomically to oxalic 
or another poisonous acid. The experimenter who drops oxalic acid into 
his culture of paramecium is so rarely encountered in nature that the 
behavioral program of this species need not provide for him. 

There are natural enemies which exploit the limited selectivity of the 
paramecium's response to acid. At certain optimal concentrations, para- 
mecia will react specifically to tactile stimulation by soft substances; they 
stop their forward movement and "anchor" by pressing gently against 
the soft obstacle. Large amoebas excrete just the right amount of C0 2 to 
make the vicinity near them attractive to ciliates, and they present to an 
approaching paramecium a surface of exactly the right, soft consistency 
to entice it to "anchor." Then the amoeba quite slowly extrudes pseu- 
dopods that surround the paramecium and finally endose and capture it. 

I once observed a paramecium which, when it was all but enclosed in 
such a net of pseudopods, went into reverse and slipped out of the trap. 
To do this it had to resort to a special motor pattern at the disposal of the 
species, that is, constricting the tip of its body sufficiently to allow it to 
enter a narrow passage and then causing the constriction to pass along 
the length of its body in a peristaltic movement. This, of course, pushes 
the body in the opposite direction, thus forcing it through the narrow 
aperture. When that particular paramecium was completely out of the 
cavity formed by the amoeba's pseudopods, that ill-advised animalcule 
shifted into forward movement and went in again, reversing the peri¬ 
staltic trick. In the next instant the amoeba had closed the trap and the 
paramecium gave a few last violent, jerking escape responses which 
made the thin membrane closing the trap bulge visibly. Then it expelled 
its trichocysts and died, which in these protozoa is visible through the 
sudden dissolution of the internal plasma structures. The whole incident 
most dramatically illustrated the danger of "erroneously" responding to 
the "wrong" stimulus configuration, a danger incurred by the limited 


II. Afferent Processes 

selectivity of all IRMs—and not only the simple ones of the lowest free- 
moving organisms. 

This danger is minimized by the "choice" of stimulus configurations 
made during the evolution of IRMs. These configurations always repre- 
sent a compromise between the greatest possible simplicity and the 
greatest attainable improbability of any external stimulus situation, other 
than the teleonomically adequate one, eliciting the specific response. A 
paragon of this combination is the classic example cited by Jakob von 
Uexküll (1909), the stinging response of the common tick (Ixodes rhi- 
zinus). The female of this species, after the last molt, can wait a very long 
time before finding a final host, any mammal. The IRM responding selec- 
tively to this host object consists of a reaction to only two key stimuli: 
first, a body temperature of roughly 37 °C and second, the smell of 
butyric acid. Furthermore, finding the proper object is helped by the 
tick's sitting patiently on low branches and allowing itself to drop down, 
when these are shaken, so that there is a good chance of its falling onto 
an animal moving below. The probability of all these conditions being 
fulfilled by anything but a mammal is negligible. On the occasion of an 
institute carnival party, one of my students composed a wonderful epic 
in hexameter describing how, on a sunny slope covered with boulders, 
a wild boar rubbed its back against one of them and set it rolling down- 
hill, and how the boulder, striking and shaking a bramble, and smelling 
of butyric acid acquired from the rubbing and secretions of the boar's 
skin, seduced a tick into trying to sting it, and how the poor tick, having 
irreparably bent its proboscis, died of depression. Nothing could better 
illustrate the strength and the weakness inherent in the IRM. 

The problem of how and when the function of selecting is performed 
by the IRM must be separately investigated for every single instance. The 
questions how and when are easiest to answer when a sensory organ is 
responding to only one single sort of stimulation to which the animal 
responds with but one single behavior pattern. As Regen (1924) has dem- 
onstrated in the common cricket (Acheta domesticus), the auditory organ 
of the female responds exclusively to the pitch of the male's courtship 
song, and the insect reacts to it by turning towards and approaching the 
source of the sound. In one of the experiments, the female jumped right 
into the loudspeaker Regen was using. Similar one-to-one linkages of 
stimulus perceived and behavior pattern released have been shown in 
the males of some mosquitoes responding to the frequencies of the 
female's wing beats. 

The same questions of how and when must obviously receive very dif- 
ferent answers if an animal possesses a large repertory of different 
behavior patterns, each of which is released selectively and teleonomi¬ 
cally by one of an equally large number of stimulus configurations, all of 
which are received by way of the same sense organ, for instance, the eye. 
In such a case one cannot avoid assuming the existence of a physiological 

1. The Innate Releasing Mechanism (IRM) 


mechansm situated between the sense organ and the motor activity that 
performs a function comparable to that of filtering incoming stimuli, and 
that permits only certain configurations to pass and to affect the locus in 
command of one particular motor pattern. Several physiologists have 
grasped the necessity of postulating such a stimulus-filtering mechanism. 
One of them was Pavlov; he very aptly named it the "detector," a term 
which is, perhaps, preferable to that of the IRM. More than half a century 
ago, the American ornithologist, F. Herrick, made the important state- 
ment, "The instincts of a species fit like lock and key," and therewith 
compared the function of impinging stimulus configurations to that of 
keys. The term key stimuli is accepted in ethology as denoting those stim¬ 
ulus configurations that, as has been explained in Two/I/5, add up to 
their effects according to Seitz's law of heterogeneous summation. 

When I said that the term "detector" is, in a way, preferable to that of 
innate releasing mechanism, I did so because the releasing of a fixed 
motor pattern does not by any means represent the only function of the 
mechanism here under discussion, just as the instinctive action, the spe- 
cies-characteristic drive action, is not the only behavioral system into 
which IRMs and motor patterns are found integrated. Still, if the term 
IRM implies that releasing is one of the most common functions of the 
mechanism thus described, the implication is not altogether misleading, 
and I propose to retain Tinbergen's term although it must be remem- 
bered that, as I shall describe later in Three/IV/3, the most common 
function of IRMs consists in switching from one link in a hierarchic 
chain of behavior to a subsequent one. Cases are known as well in which 
a typical IRM "releases" nothing but a specific inhibition! 

At the moment, very little is known about the physiologic processes 
which achieve the function essential for the IRM, that of filtering stimuli. 
The investigation conducted by J. Y. Lettvin and his co-workers on the 
retinal functions of the frog (1959) tend to show in what direction the 
solution of these problems may be expected. In the retina of the frog, 
groups of sensory elements are connected with one gaglion cell. Each 
group responds to another form of stimulus configuration reported by 
the elements. One responds only if all elements simultaneously report a 
change from light to darkness or the opposite (the on-oíf eífect). Other 
groups respond only to much more specific sensory inputs, for instance, 
to a convex contour separating light and darkness and progressing in a 
particular direction across the group of sensory elements. Each of these 
sensory elements is a member of several groups, being connected to all 
adjacent cells in the ganglion retinas. The selective response to such spe¬ 
cific configurations as "dark convex contour moving from left to right" 
is strongly reminiscent of that of the IRM in an intact animal responding 
to a key stimulus. One is indeed tempted to speculate that it should be 
easy for a more centrally situated group of nervous cells to intégrate the 
information furnished by the cells of the ganglion retinas into relevant 


II. Afferent Processes 

reports, such as "insect flying past from right to left," which could be 
directly passed on to motor systems. 

E. and P. Kuenzer investigated the response of young pygmy cichlids 
(Nannacara anómala) to the optical stimulus configurations emanating 
from their mother (1968). The young fry clearly differentiate between 
their mother and other, potentially predatory fish by swimming toward 
the former and avoiding the latter. The most effective key stimulus con¬ 
figurations eliciting approach proved to be the light and dark color pat- 
tern of the parent fish and the sideways jerking head movements made 
by the female Nannacara in the presence of her progeny—at about the 
same frequency with which a mother hen utters her clucking cali. Sur- 
prisingly, the black and white coloration of the mother was not only 
effective through the contrast between both shadings, but also through 
the contrast between these and the background: the dark markings had 
to be darker than the background, while the effect of the white markings 
was dependent on their contrast with both the dark pattern and with the 
background. Thus the light and the dark elements of the female's color 
pattern appeared to furnish sepárate pieces of information. What proved 
essential for the releasing effect were not absolute quantities of stimula- 
tion, but contrasts, in other words, relationships between impinging 
stimuli. The Kuenzers also examined the physiological conditions which 
are responsible for the releasing function within the sensory organ itself. 
On the basis of these results they succeeded in constructing "super-nor- 
mal" dummies that, with regard to their releasing effect, by far surpassed 
the natural object and were able to lure the Nannacara babies away from 
their real mother. 

When watching the prompt and teleonomically correct response 
which the normal object calis forth by way of an IRM, the naive observer 
tends to overrate the amount of information contained in its program. To 
see a newly-hatched turkey (Meleagris gallopavo) hide in cover and crouch 
at its first sight of a hawk flying over, or to see a young kestrel (Cerachneis 
tinnunculus) , on first encountering water, bathe and preen as if it had 
done these things hundreds of times, is indeed impressive. To see turkey 
chicks react in the same manner to a fat fly slowly crawling along the 
ceiling, or the young kestrel trying to bathe, using the same movements, 
on a polished marble table, is actually disappointing. Yet these errors are 
significant and were the first indications which led us to a better under- 
standing of the IRM and of the paucity, or better, the parsimony of the 
information it conveys. 

Tinbergen demonstrated the surprising simplicity of key stimuli and, 
at the same time, their additive effect in the IRM directing the gaping of 
nestling blackbirds (Turdus merula) towards the head of the parent birds 
bringing food. If one offers to the nestlings two rods at the same eleva- 
tion, they will gape at the nearer one; if one offers the rods at the same 
distance but at different elevations, they will gape at the higher one. If 

1. The Innate Releasing Mechanism (IRM) 


Figure 15. Any discontinuity in contour acts as "head." With a projecting triangle, 
it is always the upper córner which attracts gaping. (Lorenz: Studies in Animal and 
Human Behaviour, Volume II.) 

one presents two objects, for instance, cardboard dises of different size at 
equal distances as well as at equal elevations, they will gape at the 
smaller one. Tinbergen presented to the nestlings a dummy consisting 
of two cardboard dises differing in size. Offering this dummy at first with 
the smaller disc up, and then rotating the dummy slowly and thereby 
lowering the smaller disc, Tinbergen ascertained how far this could go 
until the gaping ceased to follow the smaller disc and was directed at the 
upper edge of the larger one (Figure 15). By playing the stimulus config- 
uration of "higher" against that of "smaller" Tinbergen ascertained the 
optimum relationship of the dises: the gaping of the nestlings directed 
at the smaller disc followed this the farthest downward when its diam- 
eter was one-third that of the larger disc. The relationship was indepen- 
dent of the absolute size of the model. Analogous experiments with 
regard to the configurations of "higher" and "nearer" brought analogous 
results, the first proving much more effective than the latter. The opti¬ 
mum difference between the two was not investigated. Two indentations 
in the contour of a single disc caused the nestlings to direct their gaping 
at the area between them, provided they were at the right distance from 
each other to properly simúlate a head, but they did not follow this as 
far downward as they did a more distinctly separated "head" when it 
was lowered. For her doctoral dissertation Use Prechtl investigated the 


II. Afferent Processes 



Figure 16. a and b. The schema orienting the gaping responses of young black- 
birds. With two rods at the same distance, gaping is directed at the higher (a: side 
view), while with two rods at the same height, gaping is towards the nearer ( b: 
overhead view). In c, height is offset against nearness; height wins. (Lorenz, Stud- 
ies in Animal and Human Behaviour, Volume II. Adapted from Tinbergen, "The 
Releasing and Directing Stimulus Situations of the Gaping Response in Young 
Blackbirds and Thrushes.") 

effects of further criteria which constitute properties of the real parent 
head. First, by using three-dimensional models, she showed that the con- 
figurations of "higher" and "nearer" when presented simultaneously do 
indeed have the summational effect predictable on the basis of Seitz's 
rule of heterogeneous summation. Furthermore, any structures of the 
head, particularly those which protruded in the direction of the gaping 
nestling, added to the releasing valué of the model's "head." On the 
whole, Ilse Prechtl's models showed a remarkable similarity to a very 
crude toy bird. 

All these experiments show a number of properties inherent to the 
IRM which tend to explain why the assumption of a simplified image, of 
an "innate scheme" of the object seemed promising. On the other hand, 
they show very clearly that the phyletic information is not given to the 
organism in the form of a unitary, if simplified, image of the object, but 
by a number of mutually independent responses to very simple config- 
urations—to key stimuli—whose effects add up in accordance with 
Seitz's law. 

1. The Innate Releasing Mechanism (IRM) 


Although the sum of the key stimuli to which an IRM responds does 
not, by any means, represent a unitary complex quality, the "gestalt" of 
its object, nevertheless each of the stimulus configurations that act as 
"keys" to the response can be regarded as a simple gestalt in itself. Its 
effect as a signal is never dependent on absolute stimulus data, but 
always on the perception of intervals , of differences and relationships, all 
of which can be represented by very different absolute valúes. The only 
exception is represented by mechanisms whose selectivity is dependent 
on the narrow margins of receptivity in the sense organ itself, as in 
Regen's cricket (page 156). The relatively simple stimulus configurations 
which act as "keys" to an IRM are closely akin to gestalt perceptions in 
this respect: Christian von Ehrenfels, one of the pioneers of Gestalt psy- 
chology, emphasized that "transposibility" is one of the criteria of gestalt 
perception (1890). The classic example is the recognition of a melody; 
this is quite independent of absolute pitch, of the instrument which pro¬ 
duces the sounds, and so on. Transposibility is equally characteristic of 
all key stimuli. It has already been mentioned that, for the orientation of 
gaping in nestling thrushes, the "head" has to be one-third the size 
of the "body," independently of the absolute size of the dummy. The 
contrast phenomena that, as the Kuenzers have demonstrated, are essen- 
tial for releasing following responses in young Nannacara (page 158) are 
another example. A third example is furnished by the response of male 
mouthbreeders of the species Haplochromis burtoni to the markings on the 
head of the rival. Among these, a dark bar traversing the eye represents 
one of the most effective key stimuli releasing rival fighting. This bar 
extends from the pupil of the eye obliquely downward and forward. As 
Leong showed through dummy experiments (1969), the "transposable" 
character essential for the releasing effect is not the angle between the 
bar and the horizontal, which would have been possible to assume since 
the males, in the typical broadside-on threatening posture, are positioned 
more or less horizontal, but the angle between this bar and the longitu¬ 
dinal axis of the fish's body. 

The overwhelming majority of IRMs that hitherto have been investi- 
gated are composed of several responses to key stimuli which are, in 
principie, independent of each other, and which add up to a qualitatively 
uniform effect. The rule of heterogeneous summation has prevailed in 
every case examined. Only two instances are known to me in which one 
single configuraron of stimuli proved to be effective; in both cases this 
configuraron was singularly complicated and, in this respect, gestalt- 
like. The so-called rattling attack of the jackdaw (Coleus monedula) serves 
to defend a fellow member of the species which has been taken by a 
predator. To release this behavior pattern it is necessary that an object 
that is a) black, and b) soft and dangling or fluttering, be carried by a live 
creature. I discovered the response when I happened to carry a pair of 


II. Afferent Processes 

wet black bathing trunks in my hand. As further experiments showed, 
it is quite irrelevant for the response what sort of a creature is carrying 
the soft black object; a nest-building jackdaw trying to carry a raven's 
secondary feather to its nest was furiously attacked by all the jackdaws 
within sight. An object that was rigid in itself, though black and carried 
by a live being, as for instance a black camera box carried by myself, 
elicited no response, but I had to hide from my jackdaws whenever I 
pulled out the black paper strips which, at that time, served to exchange 
the films. Whenever the birds saw these fluttering black objects in my 
hands, I was subjected to mass attack immediately. The number of such 
experiments I could conduct was limited by the aftereffects of each rat- 
tling attack; the birds were severely alarmed and remained distrustful of 
me for a long time after each such event. A greater number of triáis 
would have spoiled my chances of further observation. 

The second example of an IRM which responds to a single but com- 
plicated and gestalt-like configuration of stimuli was found by O. Drees 
in male jumping spiders (Salticidae) (1952). At a certain distance, these 
animáis respond to any small black visible object by running towards it. 
The initial behavior leading to either catching prey or to courting a 
female is identical. Which of these two behaviors will follow as the 
sequel is determined only when the male has approached the object near 
enough to discern its contours. Near enough is a distance of a few cen- 
timeters. Then the object, or a dummy connected to the substratum by 
short vertical "legs," releases the prey-catching jump. At the sight of an 
object showing long legs, which, like a spider's, are directed upward 
from the body at their base and, arching high above it, touch the sub¬ 
stratum only at a certain distance, the male will stop and begin to wave 
its palpi in species-characteristic courtship movements. 

These two IRMs, of jackdaws and of Salticidae, represent rare and spe- 
cial cases of an innate response being elicited by a single, but compli- 
cated configuration of stimuli. In view of their theoretical importance, 
these phenomena ought to be investigated further. 

2. Limits to the Functions of IRMs 

Everything said in the preceding section was intended to show that IRMs 
are unable to respond to complex qualities in the way that learned gestalt 
perception can. A practically unlimited number of single criteria can be 
integrated into a single, unambiguously perceived quality by the learn- 
ing of a "gestalt." At the age of five years my eider daughter proved able 
to identify as members of the family of rails (Rallidae) all the different 
species which at that time were exhibited in the Schónbrunn Zoo. This 
was remarkable for two reasons. One is that the rails inelude forms 

2. Limits to the Functions of IRMs 


adapted to very different biotopes: long-legged wading birds, prairie 
dwellers extremely similar to gallinaceous birds, and ducklike aquatic 
forms. The second reason is that the little girl was familiar only with the 
aquatic forms—the coot (Fúlica atra) and the moor hen (Gallínula cholo- 
ropus). When asked how she recognized all these birds as Rallidae, my 
daughter could only say that they were "somehow like a moor hen." 
Unlike the few configurations essential to an IRM, the innumerable cri- 
teria integrated in a gestalt perception cannot easily be verbalized. 

The simplicity and poverty of the information conveyed by an IRM 
can, as already mentioned, lead to "errors" and can do so even under 
natural circumstances. Geese can react to a leaf wafted along by a slight 
breeze as if it were a slowly gliding eagle. I have known incubating tur- 
key hens to roll smooth pebbles or—in one case—a tin cigarette box into 
their nests because the IRM of egg rolling responds to any object that is 
hard, smooth, and devoid of projections. 

Another weakness is inherent in the IRM for the very reason that the 
information which it contains is coded in relations and not in absolute 
valúes. As a result of this, it is possible to exaggerate certain relationships 
which act as key stimuli and to construct models whose efficiency by far 
surpasses that of the natural object. Tinbergen has demonstrated with 
gulls and Baerends with the Oystercatcher that the incubating bird pre- 
fers oversized egg dummies painted with a pattern of strong contrasts to 
its own clutch. When in an experiment involving choice the bird is 
offered such dummies side by side with its own eggs, it will promptly sit 
on the former. The presence of the supernormal object effectively pre¬ 
venís the silly bird from normal incubation. 

Another example of an IRM that can easily be misled by the exagger- 
ation of one key stimulus is the one eliciting copulatory actions in geese. 
Besides some pre-copulatory movements, the most effective stimulus 
configuraron consists of the partner's offering a large horizontal plañe 
near the surface of the water, as does the female goose when inviting the 
male to mount. The effect of this horizontal surface can be very much 
enhanced by making it larger and by offering it just a few centimeters 
below the water surface. Unintentionally the human foster parent does 
just this when going into the water with tame geese. Even juvenile geese 
not yet able to fly, including females, will try to mount the swimmer's 
back, scratching most horribly. This is done—as must be emphasized— 
by individuáis that are not sexually imprinted to humans. The informa¬ 
tion conveyed by the IRM would read when verbalized: "Large horizon¬ 
tal plañe near water surface offered by conspecific." 

In a manner of speaking, IRMs of this kind are "open to exaggeration 
in one direction"; verbalized they would read "as large as possible," "as 
rich in contrast as possible," and so on. There are a number of social par¬ 
asites which employ the method of exaggerating key stimuli, as does the 


II. Aíferent Processes 

European cuckoo, and a number of social parasites among insects. Hein- 
roth observed that a fully fledged young cuckoo which he put in an avi- 
ary was fed by birds of quite varied species, not only by adults, but also 
by juveniles that had only just become independent of being fed them- 
selves. Apparently the huge gape of the young cuckoo represents a 
supernormally strong key stimulus to the feeding behavior of many 
kinds of passerine birds. Heinroth remarks, "The feeding of a young cuc¬ 
koo is, in a manner of speaking, a vice of these birds"—a statement of 
remarkably deep insight. 

Human beings also possess a number of IRMs that are "open to one 
side" and which respond to supernormal stimuli. An American journal- 
ist, at an ethological conference, having seen G. P. Baerends's film of an 
oystercatcher trying in vain to sit on an oversized and brightly painted 
egg, exclaimed, "Why, that's the cover girl!"—which showed complete 
understanding of the phenomenon. Most measures taken by fashion to 
enhance female—and male—beauty function on the principie of exag- 
gerating key stimuli. The same is true of the dolí industry. Humans 
respond with emotions and behavior patterns of parental care to a num¬ 
ber of configurational key stimuli that can easily be analyzed—and also 
exaggerated. One of them is a high and slightly bulging forehead, a brain 
case large in proportion to the face and the visceral cranium, large eyes, 
rounded cheeks, short and stubby limbs, and a rounded fat body. Addi- 
tional key stimuli are uncertain, stumbling movements. A puppy which 
can keep to its intended direction as long as it walks, but deviates from 
it the moment it tries to gallop, is surprisingly sweet. It is typical for the 
unreflecting nature of the IRM that we feel moved to tenderness even 
by adult animáis provided they possess the characters just mentioned. 
Such creatures are felt to be engaging, sweet, or appealing; in Germán 
their species ñames often end in the diminutive suffix -chen (Figure 17). 

T. Tugenhat made an interesting series of experiments with humans, 
showing them models in which the configurational key stimuli were 
depicted in different degrees of exaggeration. Her results were somewhat 
contradictory and I believe this was caused by the kind of instruction 
given to her experimental subjects: she instructed them to choose the 
model that was "most babylike." B. Hückstedt, who had no bias against 
introspection and the recognition of emotions, made the same sort of 
experiments, but she told her subjects to choose the model they would 
prefer to cuddle. Her results were quite unambiguous and also demon- 
strated the validity of the law of heterogeneous summation. As an inter¬ 
esting side result, she was able to explain why the teddybear and Bambi 
can compete with human effigies as effectively as they do. The Kewpie 
dolí represents the máximum possible exaggeration of the proportions 
between cranium and face which our perception can tolérate without 
switching our response from the sweet baby to that elicited by the eerie 

2. Limits to the Functions of IRMs 


Figure 17. The releasing "schema" for human parental care responses. Left: head 
proportions perceived as "lovable" (child, jerboa, Pekinese dog, robín). Right: 
related heads which do not elicit the parental drive (man, haré, hound, golden 
oriole). (Lorenz: Studies in Animal and Human Behavior, Volume II.) 

monster. The eerie-monster response is elicited whenever a very well- 
known gestalt perception is disturbed by one or several unknown new 
characters. The conventional pictures of devils and ghosts illustrate this 
phenomenon. The eerie-monster phenomenon is the more potent, the 
better the distorted gestalt is known. For this reason the proportions of 
cerebral to visceral cranium can be exaggerated to a much higher degree 
in the picture or in the model of an animal without causing revulsión 
than it can in the human picture or model. An exaggerated baby donkey 
was about the most strongly supernormal dummy B. Hückstedt could 
devise. Of course, the dolí industry has for a long time been aware of all 
this and has successfully exploited this knowledge. 

The art of cooking, the competition of chefs catering for the most 
sophisticated gourmets, long ago inspired the invention of supernormal 


II. Afferent Processes 

food stuffs, much to the detriment of civilized humanity. For our paleo- 
lithic ancestors, hungry as they were much of the time, it was certainly 
sound strategy to follow the instructions of IRMs telling them what 
foods to choose: they should contain as much fat as possible, as much 
sugar as possible, and as little roughage as possible. Being "open on one 
side," these key stimuli led to an extremely unhealthy preference for 
supernormal objects. White flour is causing severe and obstínate consti¬ 
paron in millions of civilized people; chocolate combines all the three 
key stimuli mentiond above, being devoid of indigestible fiber and con- 
sisting of fat and sugar. Even the most complete insight into the work- 
ings of our IRMs does not make it easy to avoid suicide by overeating. 

3. IRM and the Releaser 

If, in the interest of survival, an animal has to react innately and selec- 
tively to a certain object in its environment, the máximum possible 
adaptedness is reached when the IRM responds with the greatest possi¬ 
ble selectivity to key stimuli which the object is emitting. It is beyond 
the powers of any organism's evolution to endow the object of an innate 
response with characters rendering it more unambiguously recognizable. 
This becomes possible only if the acting subject and the object of its activ- 
ity are members of the same species. The pike, figuratively speaking, is 
not able to affix a little flag to the silvery minnow that will then release 
its snapping, and thus keep itself from snapping at any silvery lure. But 
a species of songbird, during its evolution, is perfectly able to attach col- 
orful signáis to the gape of its nestlings and thus ensure that only its 
own progeny will be fed by it. Structures and motor patterns—or com- 
binations of both, as in most cases—that are thus evolved in the Service 
of emitting key stimuli are called releasers. As a stickler for nomenclature 
I must here cali attention to a current misuse of this term: it is often mis- 
leadingly used to denote releasing stimulation in general, more or less 
in the sense in which the term 'key stimulus' would be correct. In its 
proper sense, the term 'releaser' connotes a structure or a movement, 
most often a combination of both, which has evolved in the Service of 
sending a signal, that is to say, of emitting key stimuli. 

It has been proposed, with some justification, that the adjective 'social' 
be joined to the term 'releaser' because, in fact, the vast majority of key 
stimuli emitted by releasers are addressed to fellow members of the same 
species. There are, however, many stimulus-sending organs and move- 
ments which have been evolved in order to influence animáis of other 
species. Many insects have convergently evolved means to frighten away 
predators: the beautiful eye spots on the wings of moths and butterflies, 
as well as on the front legs of many mantidae, have been demonstrated 

3. IRM and the Releaser 


to perform this function. An octopus uses the enormous motility of its 
body as well as of its chromatophores to make appear, with really fright- 
ening suddenness, a huge pair of eyes on the surface facing an approach- 
ing predator; squids ( Sepiotheutis) "paint" magic eyes on their lateral fins. 
For obvious reasons, the frightening effect of such eye spots is the greater 
the farther they are apart, in other words, the larger the head of the sim- 
ulated predator is made to appear. 

Some releasers have evolved to influence not the predator but the 
prey. The large North American snapping turtle (Macroclemmys telmincki) 
possesses, on the tip of its tongue, a very convincing imitation of a worm 
that, by its wriggling, entices fish to swim right into the turtle's mouth. 
Angler fish (Lophiidae) also possess lures attached to the first rays of 
their spiny dorsal fins, situated near the front of their heads. In the split- 
lure anglerfish (Phrynelox scaper), this lure consists of a wormlike appen- 
dage which wriggles exactly like a worm. 

The vast majority of releasers, however, serve intraspecific communi- 
cation. It is hardly an exaggeration to say that every striking or "showy" 
color pattern or structure found in a vertébrate, as well as any loud and 
regular sound utterance, or any regular, complicated, and rhythmically 
performed movement, functions as a releaser. On observing a striking or 
colorful structure, it is very often possible to predict without any obser¬ 
varon of behavior the way in which this structure will be presented to 
the conspecific. Tinbergen and I happened simultaneously to obtain a 
species of cichlid (Cichlosoma meeki) new to both of us, which had eye 
spots not, as many other members of this group, on the operculars but 
on the gilí membranes. Our letters, with drawings, in which both of us 
correctly predicted the form of frontal display peculiar to this species, 
crossed en route. 

A few further examples can suffice to illustrate the principie. Morpho- 
logical releasers often serve to make a motor pattern more conspicuous; 
the motor pattern itself must not be changed in the interest of its teleon- 
omic function. Many birds, when taking off, disclose color patterns 
which remain hidden under their wings while they are grounded. The 
wing speculum of ducks, the white rump of geese, many finches, and 
others, and the brightly colored lateral tail feathers of the budgerigar 
(Melopsittacus undulatus) are examples. Morphological releasers totally 
unconnected with any particular motor pattern are not very common. 
The only examples I can cite offhand are the color patterns permanently 
shown by certain coral reef fish. These—unlike the colorful patterns of 
many freshwater fish—are shown constantly and do not alter with 
changing moods. Their primary function is the release of aggression 
against fellow members of the species, in other words, to ensure aggres- 
sivity being strictly intraspecific, in still other words, to avoid unneces- 
sary fighting with fish that are not food competitors. The multiplicity of 


II. Afferent Processes 

ecological niches concentrated on a reef, and the consequent number of 
species and of individuáis crowding on it, have exerted a selection pres- 
sure under which the incredible gamut of coloration has evolved. 

In the case of social releasers, the stimulus-emitting apparatus and the 
stimulus-receiving apparatus are in a position to exert pressure on one 
another. The releaser obviously "caters for" the particular properties and 
limitations inherent in the IRM, and the IRM can conform to the ever- 
growing unambiguity of the key stimuli sent to it. In this way a type of 
communication can evolve that, functionally, is closely analogous to the 
use of symbols. Motor patterns that have undergone a high differentiation 
in the Service of their releasing function have been described by many 
observers—who were far from naive—as ritualized symbolic actions. 

As mentioned, the great majority of releasers consists of a combination 
of a motor pattern acting as a signal and a morphological structure 
enhancing the signal's effect. Very often a comparative study of many 
closely related species reveáis the fact that motor patterns are phylo- 
gentically older than the structures serving to make them more conspic- 
uous. The courtship movements of male dabbling ducks (Anatini) fur- 
nish many good examples of this. A homologous movement is often very 
similar, in many species, while the patterns and structures of the feathers 
that make the movement optically more effective are quite different, 
although they obviously evolved under the same selection pressure. We 
know more about the phylogeny of this sort of releasing motor pattern 
than we do about any other kind of genetically programmed pattern of 
behavior, because we know quite a number of beaütiful series of differ- 
entation and, what is more, we know in which direction they should be 

Man, too, is in possession of a number of true releasers, as Charles Dar- 
win was the first to point out. Certain expressive movements, such as the 
smile—studied extensively by René Spitz and Irenáus Eibl-Eibesfeldt— 
have been demonstrated to be genetically fixed with respect to their 
motor coordinations, as well as to the responses evoked in the addressee. 
Like the "póster colors" of coral fish, some bodily structures of humans 
can be shown to have a releasing effect which is more or less indepen- 
dent of motor patterns. The distribution of fatty tissue over the body sur- 
face can cause configurations which demonstrably have the effect of 
emitting key stimuli. One example is the corpus adiposum buccae, the 
bunch of fatty tissue in the cheeks of small children. Farfetched assump- 
tions have been made to explain its presence, for instance, that it has 
played a role in the mechanism of sucking. If so, how it is possible for 
all other primates to manage without the organ remains to be explained. 
On the other hand, its function in helping elicit responses of parental 
care has been demonstrated. Unlike that of the corpus adiposum buccae, 
the releasing effect of the specific distribution of fatty tissue on the 
female body does not seem to have struck any observer as problematic. 

3. IRM and the Releaser 


Releasers of a very special kind have evolved whenever the advantage 
of being mistaken for another species has exerted a selection pressure. 
The best known and perhaps also the most common examples are fur- 
nished by many "imitating" or "mimicking" species which are, in some 
way, protected by poisons, repellent taste, or by other means. Most of the 
species thus protected are colored and marked in a strikingly aposematic 
way. These marks have the effect of warning off predators either because 
they possess an innate aversión to them or because they have had indi¬ 
vidual experience of the repellent agent and have associated this with 
coloration. Aposematic colors emit stimuli that may be received by an 
IRM or, in other cases, by a learned response. Any unprotected and tol- 
erably similar species may, of course, enjoy the advantages of this pro- 
tection by evolving toward even more similar coloration. 

One particular kind of mimicry consists of imitating certain social 
releasers of another species and thereby profiting from being treated in 
the same way as a fellow member of the species. The parasitic whydah 
birds (Viduini), which lay their eggs in the nests of waxbills (Estreldi- 
nae), imitate the juvenile plumage and particularly the gape of their 
hosts so exactly that they deceived even J. Nicolai (1970) who, thinking 
he had brought home nestling whydahs, later found that two of his birds 
belonged to the species of their host. The closeness of the mimicry in the 
young both with regard to morphology and to behavior is necessary 
because the whydahs, unlike the European cuckoo, do not destroy the 
young of their hosts. As Nicolai could convincingly show, the brood par- 
asitism of whydahs is phyletically rather oíd; their speciation proceeded 
parallel to that of their hosts. In consequence of this, there are some spe¬ 
cies which belong to the same genus and are hard to distinguish as 
adults, while their neonates and juveniles differ considerably, being 
adapted to mimic the morphology and behavior of different host species. 
These species have always been considered by older ornithologists as 
subspecies, if not as mere races. As Nicolai could demónstrate, they 
deserve without question the taxonomic rank of a "good" species, since 
they are sympatric (occur in the same area) and yet never hybridize. To 
do so would be disastrous for the offspring because the young hybrid, 
being adapted neither to one ñor to the other host species, would have 
no chance of survival. Under this particular selection pressure, the why¬ 
dah birds have evolved a barrier against hybridization which, as Nicolai 
has demonstrated, is based on learning processes of the imprinting type, 
which will be discussed in detail in Three/III/6. 

In a very peculiar case of mimicry, releasers can simúlate organs of the 
same species. W. Wickler, who has investigated this phenomenon thor- 
oughly, speaks of "automimesis." When the first specimen of the mouth- 
breeder Haplochromis burtoni had just arrived in Seewiesen, an American 
guest, John Burchard, Jr., pointed out to me the strikingly shaded orange 
spots on the anal fin of the males and offered a bet that he could tell me 


II. Afferent Processes 

their function. I cut him short by saying that the spots were imitation 
eggs, and that the female, after spawning, would snap at them while the 
male discharged his sperm—which was exactly what Burchard had sus- 
pected. Wickler proved the correctness of this rather obvious assumption 
and made an important comparative study of the evolution of the "egg 
dummies" in African mouthbreeders. It is interesting that so small a 
selection pressure should produce such a very special adaptation of mor- 
phology and behavior. The only advantage gained is that the eggs can 
be taken into the mouth of the mothers at once and then inseminated 
while she snaps at the egg dummies, instead of remaining exposed for a 
number of seconds until the male has inseminated them. 

Wickler has drawn attention to some other cases of automimesis: in 
several species of baboons and of vervet monkeys, the male carries on 
his rump structures and colors imitating the genitalia of the estrous 
female. They are presented during the appeasement gesture to the dom- 
inant male. In the gellada baboon (Theopithecus gellada), a detailed imi¬ 
tation of the rear aspect of the female is shown on the front of the male's 
chest, and this is also presented during the gesture of submission. An 
interesting detail is that the structures imitated and the structures imi¬ 
tating need not, by any means, be homologous. The red callosities on the 
rump of male babooons imitate the labia minora of the female; in some 
vervet monkeys red hairs cover regions of the male's body which, in the 
female, show a strongly vasculated skin. 

Wickler has pointed out that a comparative study of mimicry furnishes 
particularly good objects for investigating the evolution of releasers. 
Since mimicry is caused by a selection pressure exerted by the IRM of 
another species, it is a one-sided process of adaptation, while in all social 
releasers the stimulus-emitting and the stimulus-receiving apparatus 
evolve in a complicated process of mutual adaptation. 

4. An Important Rule of Thumb 

Very different though the releasers are that have evolved among all the 
phyla of higher animáis, they have certain properties in common. This 
can easily be explained on the basis of what has been said in the preced- 
ing section about the functional limitations of the IRM. All releasers emit 
relatively simple spatial and temporal stimulus configurations and, as far 
as they are visual, they employ simple, that is to say, nearly puré spectral 
colors. In other words, they have all been evolved under the selection 
pressure mentioned in the preceding section, which tends to give them 
the greatest possible simplicity and, at the same time, the greatest possi- 
ble unambiguity. It is characteristic of releasers that they can be 
described in comparatively few and simple words. This fact can easily be 
made evident by reading, in any ornithological book, the description of 

4. An Important Rule of Thumb 


the male and the female of the species, in which the male has nuptial 
plumage, while the female is cryptically colored. A mallard drake with 
its green head, white neck band, brown breast, et cetera, can be described 
with very few words, while the description of the female, in order to 
afford equally detailed information, must fill many pages. 

As has been mentioned on page 161, the IRM cannot do what is so 
easily done by our learned gestalt perception, that is, respond selectively 
to complex qualities. It is an extremely reliable rule of thumb that an 
IRM can be assumed to be at work whenever an organism is "taken in" 
by a very simple dummy or model. Conversely, if the attempt to elicit a 
certain response by a dummy fails, and it proves necessary to simúlate a 
biologically relevant stimulus situation in all its details in order to release 
a response, or if even this proves to be impossible, the assumption is jus- 
tified that the organism has learned to respond to a complex quality. This 
rule was brought home to me in a most unforgettable way by the follow- 
ing experiments made by Alfred Seitz. When he had performed his now 
classic experiments on releasing fighting in male Astatotilapia, he next 
attempted to elicit, by the same method, the male's responses of court- 
ship. He was quite unsuccessful with simple models. Then he proceeded 
to build increasingly complicated imitations of the female fish, finally 
arriving at models made of semi-transparent paraffin fitted with celluloid 
fins and rendered silvery by the application of powdered aluminum. 
These models might easily have deceived a human observer when they 
were suspended in the aquarium by the thinnest of nylon threads. They 
did not, however, deceive the male fish, not even when these were suf- 
fering from a considerable lowering of threshold. 

Having more confidence in my theories than I had myself, Seitz con- 
cluded that the male Astatotilapia's response to the female must be con- 
ditioned and based on learned gestalt perception, and he decided forth- 
with to rear some males in complete isolation, depriving them of any 
opportunity to learn what a female looked like. The deprivation experi- 
ment was not at all easy with a mouthbreeding species whose eggs have 
to be continously whirled about as they normally are by the breathing 
movements of the mother. Seitz's patience did not flinch before the task 
of constructing an egg-whirling apparatus consisting of a number of 
upward directed jets of warm, aerated water on which the eggs were kept 
dancing. When he had finally succeeded in rearing to full maturity five 
males which had never seen their own kind, Seitz invited me to watch 
the crucial experiment. In order to avoid any possible conditioning influ- 
encing later experiments, Seitz began with the most simple dummy pos¬ 
sible, a grey sphere of plasticine stuck onto a thin glass rod. When this 
dummy was shoved into the tank containing one of the males and was 
gingerly brought near the fish, the latter, which up to then had sat quies- 
cent and cryptically colored within the cover provided by a plant, 
quickly blushed into full nuptial coloration, erected the median fins. 


II. Afferent Processes 

oriented broadside to the model, and burst into a bout of courtship move- 
ments of the utmost intensity. It bears witness to a commendable distrust 
in my own theories that I was utterly surprised by this. The rule of 
thumb just mentioned had been derived mainly from the observation of 
birds, and I had not really believed that the learning of complex gestalt 
perception could play such an important part in the behavior of a fish. 

From all that we know about releasers, it is safe to conclude that it is 
not possible for the evolution of sensory-neural organizations to con- 
struct a receptor mechanism which responds selectively to a complex 
quality composed of a great number of stimulus data. If it were, there 
would be no selection pressure causing the evolution of all those innu¬ 
merable stimulus-sending contrivances which gladden our eyes and our 
ears; there would not be any puré colors of flowers, ñor beautiful colors 
of coral-reef fish or birds, ñor would there be any puré notes in bird 

The IRM obviously cannot do what the learning of gestalt perception 
finds so easy, that is, to characterize a complicated multiplicity of sensory 
input so that it becomes umambiguously recognizable. The surprising 
accomplishment of gestalt perception, of which an example was men¬ 
tioned in II/2, is by no means granted exclusively to human beings. A 
tiny greylag gosling, not much more than a day oíd, has already learned 
to recognize its parents among a hundred other geese by the physiog- 
nomy of their faces as well as the individual qualities of their voices. 
Curiously enough, the same parts of the face, that is, those areas around 
the eyes and nose, are essential for recognition in both humans and 

The diíference between the functions of the IRM and the learned per¬ 
ception of gestalt qualities is indeed impressive, but nonetheless the dif- 
ference is only quantitative and concerns only degrees of complication. 
Doubtlessly the same elementary processes of perception take part in 
both functions and dictate one fundamental similarity: the information 
received is always couched in terms of relations, or intervals between two 
or more stimuli, and not in the absolute valúes of single stimulus data. 
For reasons obvious to the cyberneticist, it is much easier to build a recep¬ 
tor apparatus responding selectively to signáis consisting of precise and 
simple relationships between stimulus data, preferably to configurations 
that can be expressed in whole numbers, in other words, that are "har- 
monious." On the afferent side of the central nervous system, there 
seems to be an interaction between elementary receptor processes that is 
somewhat akin to the interaction which, on the motor side, takes place 
between processes of endogeneous production of impulses, and which 
also results in products whose stability is dependent on the degree of 
harmony attained. Harmony is generally improbable in itself, and 
thereby contributes to the unambiguity of any configuration perceived. 
The Germán psychologist, Félix Krüger, termed this phenomenon. 

5. IRMs Rendered More Selective by Learning 


which is common to a gestalt perception, the Pragnanztendenz, the trend 
toward terseness in perception (1948). The urgent need which all percep- 
tual processes have for harmonious configurations of stimuli exerts a 
selection pressure not only on the evolution of IRMs, but equally in the 
historie development of human signáis and, last but not least, of human 
art. Puré spectral colors, puré sounds in harmonious combinations, geo- 
metrical forms easy to express in whole numbers, rhythmically regular 
temporal sequences and so on, are to be found in releasers as well as in 
man-made signáis and in art. It is, in fact, this property common to all 
perception which has caused many organisms to become so beautiful. 

It is perfectly conceivable that there might be releasers, or stimulus- 
emitting organizations, the signáis of which are not addressed to an IRM 
but are received by learned perceptions, as are the color patterns of our 
flags. The functions of perception can certainly cause the production of 
signáis catering to the properties just described. While there is an abun- 
dance of man-made signáis whose properties are clearly dictated by the 
Pragnanztendenz of human gestalt perception, we know only a few exam- 
ples of phylogenetically programmed stimulus emitters without a cor- 
responding IRM, in other words, a releaser the response to which must 
be learned. Neonate greylags at first do not respond to the cali note of 
their species, and when they do, it is only the cali note of their parents 
which evokes a response. As Otto von Frisch demonstrated, newly 
hatched curlews (Numenius arquatus) do not respond to their parents' 
warning cali until they have heard it one or two times in connection 
with the optical perception of a bird silhouetted against the sky. S. Sjo- 
lander (oral communication, 1977) demonstrated that the zebra finch's 
(Taeniopygia castanotis) response to the red color of the conspecific's bilí 
was not innate, as had been supposed. He succeeded in imprinting 
young birds on foster parents with a bright green bilí. It is doubtful, 
however, if they could have been imprinted on a grey or a white bilí; 
the verbalized information of the IRM might read just "bilí of very strik- 
ing color." 

5. IRMs Rendered More Selective by Learning 

The Viennese zoologist, Otto Storch, was the first to draw attention to 
the fact that adaptive modifications concerning the receptor side of 
behavior are found on much lower evolutionary levels of the animal 
kingdom than are modifications of motor activities (1949). He distin- 
guished Erzverbs-Rezeptorik and Erwerbs-Motorik, which may be translated 
as receptor and motor learning. One of the most common learning pro¬ 
cesses in general, and one of the most basic ones, consists of rendering 
an IRM more selective by learning: further conditions for the release of 
the behavior pattern concerned are added to those already coded in the 


II. Afferent Processes 

IRM. Like all true learning processes, this represents an adaptive modifi- 
cation of behavior and will be dealt with in detail in Three/II/2. How- 
ever, the adaptive modification of the stimulus-filtering apparatus is so 
closely interwoven with the function of the IRM that it must be dis- 
cussed in this section. 

The IRM eliciting in our male " Astatotilapia" the motor patterns of 
courtship responds to very few and simple key stimuli. As Seitz dem- 
onstrated, any object that is approximately the size of a conspecific and 
that slowly approaches the male when he is located in his territory 
instantly releases his courtship movements, which consist of a ritualiza- 
tion of the motor patterns by which a "nest" is swept out for spawning. 
When, at this signal, the object neither hinches ñor approaches any 
closer, the male proceeds to the next step of spawning behavior: he turns 
nestwards with a ritualized movement, overaccentuating the undualtion 
of body and of tail fin. At this, the courtship-eliciting object must swim 
forward and, arriving over the shallow excavation representing the nest, 
it must join the male in a circular movement. All of the actions of this 
sequence can be easily released by means of a dummy if the male has 
been deprived of previous experience. After circling, the male will 
approach the dummy's underside, and at this moment his behavior will 
break off because olfactory stimulation emanating from the female is nec- 
essary to release the actions which normally follow. During the circular 
movement, the dummy must not at any moment present a fíat surface to 
the male; to this he instantly reacts as he would to the broadside-on dis- 
play of another male and starts fighting. While circling around over the 
nest, the female presents to the male a concave side; this, however, need 
not be imitated in the dummy experiment, only the presentation of a 
plañe must be avoided. A spherical model suífices to keep the circling 
going, showing that no information regarding the body form of the 
female is contained in the key stimuli. 

All this is true only of a fish reared in isolation. In a normally reared 
fish, all the key stimuli mentioned must emanate from a fellow member 
of the species. The information programmed in the IRM could be ver- 
balized thus: "Any conspecific, slowly approaching, passively tolerating 
courtship movements, then following to the nest and joining in circling 
it, is a ripe female." What a conspecific looks like is not contained in this 
innate instruction, and this gap must be filled in by a complex gestalt 
perception which the individual must learn to recognize during the 
course of its individual life. The special learning process which accom- 
plishes this will be discussed in Three/III/3. 

In higher vertebrates, such as teleost fish, birds, or mammals, it is dif- 
ficult to find IRMs which are not made more selective by individual 
learning of this type, but exceptions do exist. The selectivity of the IRM 
releasing the gobbling response in the turkey cock does not increase by 
learning, as M. Schleidt has demonstrated. More usually, IRMs not mod- 

5. IRMs Rendered More Selective by Learning 


ifiable by learning are found in invertebrates, particularly in insects and 
spiders (Arachnidae), and primarily in connection with activities that are 
performed only once, or only a very few times during an individual's 

It is a reliable rule of thumb that the function of an IRM is involved 
whenever it is possible to elicit some innate behavior pattern by a simple 
dummy, but this rule must not be inverted. As mentioned, the releasing 
mechanism is often made much more selective by association with a com- 
plicated gestalt perception, so much so that it has become quite impos- 
sible to release it even with the most sophisticated model—as has been 
shown for the courtship patterns of male " Astatotilapia." The impossibil- 
ity of evoking a response by means of dummies does not, however, imply 
that the IRM does not still play a part, ñor that the original key stimuli 
have become dispensable—as is the case in some other kinds of condi- 
tioning. The effects of key stimuli and those of conditioned stimulus con- 
figurations are added to one another, but their interdependence poses a 
problem which has not been sufficiently investigated. It is all too easy to 
offer, quite unintentionally, supernormal stimulation in the dummy. It 
is always possible that an excessively effective dummy more than com- 
pensates for the absence of conditioned stimuli. The observation that the 
copulation responses of greylag geese can be released by swimming peo- 
pie does not justify the conclusión that conditioned stimuli take no part 
in normal copulation. 

Chapter III 

The Problem of the "Stimulus" 

1. All-Embracing Conceptualizations 

The term "stimulus" is often associated with an extremely broad concept 
subsuming almost every external influence that could elicit an observa¬ 
ble response from any organic system—from a protozoan, from a whole 
multicellular organism, from a single nerve cell or even from an isolated 
neurite. We describe as a "key stimulus" that complex configuration of 
many stimuli to which the IRM of a higher animal responds selectively. 
A key stimulus elicits a very special and teleonomic response. We make 
this descriptive distinction in spite of our awareness that the selectivity 
is due to a most complex "filtering" mechanism through which any stim- 
ulation must pass—a filtering mechanism which evaluates a multiplicity 
of stimuli and reports nothing but a single, reliable signal to the higher 
loci within the animal's central nervous system. At the same time we 
speak of a "stimulus" when we cause a single, quantifiable electric shock 
of a few millivolts to excite a single neuron, or even an isolated nerve 
fiber, and occasionally the term "stimulus" is used even when an unspe- 
cific neural response is released by some change in the environment not 
"provided for" by phyletic adaptation—when, for example, a deficiency 
of calcium ions in the blood or in the nutrient solution causes some neu- 
rons to fire spontaneously. 

2. Stable and Spontaneously Active Nervous Elements 

On the comparatively low integrational level of the single neuron, 
everything causing the membrane to change its ionic conductances. 

2. Stable and Spontaneously Active Nervous Elements 


either up or down, therefore to result in either depolarization or hyper- 
polarization, or contributing to its readiness to do so, can be regarded as 
a stimulus. The now obsolete expression, a "breakdown" of membrane 
potential, that we once used, is rather metaphoric; what happens at the 
moment of nerve discharge is an alteration in the sign of electric charge 
on either side of the membrane: on its outside, the positive charge is 
transformed into a much weaker negative one, while the opposite 
change takes place on the inner side. This process can lead to the "firing" 
of the cell, although several gradational processes besides firing are 
known. Receptor cells in particular usually have gradated receptor 
potentials. Thus a signal is emitted that is addressed to one or more other 
neurons; the change of potential spreads from the cell membrane along 
that of the neuron and induces an identical process in the neuron 
addressed, provided the latter is, at the moment, in a sufficient State of 

At this point we meet with a conceptual difficulty to which Kenneth 
Roeder has drawn attention (1955). He says: 

The excitability of a nerve cannot be defined in physicochemical terms. The 
only way in which it can be made manifest and measured is by determining 
the minimum energy change, within a certain time interval and in a certain 
direction, that must be applied before the nerve will discharge an impulse. 

If an electric stimulus is used, the energy change is expressed in terms of 
electric current and the direction in terms of potential sign. Thus, we can 
say that the excitability or its reciprocal, the stability, of a nerve is propor- 
tional to the energy needed to abolish it, but we are unable to define it as a 
continuous property of the living tissue. Analogous reasoning is used in 
describing the stability of a building or other structure. In this case stability 
is expressed in terms of the forcé (pounds per square foot, wind velocity) 
that is just sufficient to cause its collapse, as compared with the forcé to 
which it is exposed under normal operating conditions. 

The relationship between the excitability changes in a stable nerve and 
those in a spontaneously active nerve is depicted in the lower graph of [Fig¬ 
ure 14]. The vertical axis represents excitability, which in practice would be 
measured in units of stimulus strength. The lower horizontal line represents 
the rest excitability of a relatively stable nerve such as the sciatic nerve; it 
corresponds to the normal operating load in the building analogy. The 
upper horizontal line represents the threshold excitability or the load at col¬ 
lapse in the building analogy. The solid curve shows the sequence of excit¬ 
ability changes that follow exposure of a stable nerve to a momentary stim¬ 
ulus of the dimensión of It can be seen that the excitability (or instability) 
increases rapidly from the rest level until threshold excitability is reached. 

In the building analogy this is the moment of collapse; in the nerve it is the 
moment of propagation of the impulse, the outward sign of which is the all 
or none electric change or action potential (upper graph). During the 
impulse the excitability of the nerve drops to zero, the absolute refractory 
period; in the building analogy collapse is at this moment complete. From 


III. The Problem of the "Stimulus' 

this point the nerve departs from the building analogy, since its metabolism 
enables it to return to its former State. The subsequent course of events varies 
in different nerves, but in large vertébrate A fibers excitability not only 
returns but overshoots the resting level, leading to a phase of supernormal 
excitability. As is shown in the solid curve, this is followed by a relatively 
prolonged subnormal phase of excitability and eventual return to the rest 

The supernormal phase of excitability is of particular interest, since it pro¬ 
vides a link between the stable and spontaneously active States of nerve. 
There is considerable variation in the relative magnitude of the supernormal 
phase, or, in other words, in the relative valúes of resting and threshold 
excitability. If the supernormal phase is relatively large or the difference 
between resting and threshold excitability is small, then the sequence of 
excitability changes must follow the dashed curve. As the rising excitability 
reaches the threshold level during recovery, the fiber becomes self-exciting, 
and impulses follow one another at regular intervals ([Figure 14], upper 
graph). Each impulse is accompanied by absolute refractoriness and fol¬ 
lowed by relative refractoriness and supernormal excitability rising to the 
threshold level. Thus, a single stimulus may initiate a sequence of impulses 
that continúes subject only to the counter influences of adaptation and 

In a sense, this activity could be considered as spontaneous activity, since 
the original stimulus that triggered a long succession of impulses could read- 
ily be overlooked. Strictly speaking, it is repetitive activity. True sponta¬ 
neous activity can be said to arise when the valúes for rest and threshold 
excitability (horizontal lines) coincide. At this point the nerve becomes com- 
pletely unstable, and a similar succession of impulses begins without the 
need for any external triggering stimulus. 

The suggestion by Pumphrey . . . that a spontaneously active fiber lacks a 
finite threshold is also illustrated in [Figure 14]. A stimulus, S 2 , may be inter- 
polated at any instant during the excitability cycle of the fiber. If the stim¬ 
ulus occurs at the instant illustrated, it must be of the relative dimensions of 
the double arrow in order to bring the fiber to threshold within the utili¬ 
zaron time. This leads to a propagated impulse and accompanying refrac¬ 
toriness somewhat earlier than it would have occurred spontaneously. The 
later its appearance in the excitability cycle, the smaller the critical size of 
S 2 , and the smaller the effect that it would have on the frequency of the 
rhythmic discharge. On the other hand, a stimulus of the dimensions of S 2 
would be completely inadequate if it occurred at the instant of Therefore, 
the sensitivity of the spontaneously active nerve element, frequency-mod- 
ulated as it is by stimuli of all dimensions, is limited only by the capacity of 
central nervous mechanisms to detect small changes in the frequency of its 
rhythmic discharge. 

Pumphrey's suggestion also gives a convincing explanation for the fact 
that some kinds of sensory cells have proved to be spontaneously active 
elements, their signáis consisting in modulations of frequency and not 
in a single discharge. 

3. Analogous Phenomena in Integrated Neural Systems 


3. Analogous Phenomena in Integrated Neural Systems 

What has been said in the preceding section about stable and sponta- 
neously active elements, as well as what has been said about the varying 
relationship between rest excitability and threshold, abolishes the sharp 
borderline between those stimuli which increase the rest excitability and 
those which directly trigger a discharge. At least as far as the activity of 
single neural elements is concerned, any influence which exerts a 
"tonic," long-term stress on the membrane potential increases the rest 
excitability and diminishes the stability of the element. A question 
important to ethological research is to what extent the relationship 
between stimulus strength, excitability, and threshold is analogous to 
that existing between the complicated stimulus configuration acting as 
the "key stimulus" on an IRM and the fluctuating readiness of the organ- 
ism to perform the motor pattern to-be-released. 

As has been emphasized in Two/I/14, the attempt to explain the func- 
tional properties of any integrated system on the basis of the functional 
properties of its elements, is an undertaking fraught with danger. The 
central nervous system has a deceptive proclivity for accomplishing anal¬ 
ogous functions by entirely different means and on entirely different lev¬ 
éis of integration, but in ways so similar that even the wary investigator 
can be misled into believing them to be identical. Nevertheless, the anal- 
ogy of functions that can be physiologically investigated justifies the for- 
mulation of a hypothesis, and what has been said in the preceding sec¬ 
tion does justify just as much as what was said in Two/I/14 about the 
relationship between spontaneous generation of impulses, central coor- 
dination and the properties of the fixed motor pattern. 

According to the laws of heterogeneous summation (Two/I/14), the 
key stimuli to which an IRM responds produce, on a higher level of 
integration, strictly analogous results, as do stimuli of varying dimen- 
sions with nervous elements: those too weak to cause an immediate dis¬ 
charge still serve to increase the State of readiness for a certain motor 
pattern, much as electric stimulation stresses membrane potential and 
thus facilitates its final breakdown or switch. Even the effect of those 
complex configurations of stimuli which must pass through a highly 
selective "filter" in order to exert a specific influence on equally specific 
efferent processes is dependent not only on the quantitative dimensions 
of the key stimulus, but also on the State of readiness prevailing, at the 
moment, in the stimulated system. It depends on this State of readiness 
whether the key stimulus immediately releases the specific efferent 
response, or whether it merely increases the readiness to discharge it by 
raising the present level of action specific potential, in other words, by 
increasing its valué so that the threshold valué is approached. The com¬ 
plicated afferent sector of this kind of process "behaves," in many cases, 
in a manner strictly analogous to that of the membrane potential of a 


III. The Problem of the "Stimulus' 

single nervous element. Neither the absolute valué of rest excitability 
ñor that of the threshold can be measured directly any more than they 
can in the A fibers investigated by Roeder. The only valué that can be 
measured is the distance between rest excitability and the threshold rep- 
resented in Roeder's diagram by the varying dimensions of S. 

My oíd, much ridiculed "phycho-hydraulic" thought model, repre- 
sented in Figure 18a, shows the steadily rising level of ASP (action-spe- 
cific potential; see below) during the quiescence of a fixed motor pattern 
and also the effect of external stimulation opening the discharge valve. 
However, this model misleadingly implies a qualitative difference 
between those stimuli which fill up the reservoir of ASP and those which 
finally release the motor pattern; it fails to show that the charging and 
releasing stimuli differ only in quantity. It is a well-known fact that stim¬ 
uli, each of which alone is too weak to cali forth a motor response, will 
finally do so if applied repeatedly in a long and persevering sequence. 
This is called "summation of stimuli," and is analogous to the loading of 
the cell membrane at the elementary level. 

It is Kenneth Roeder's well-founded assumption that the increase in 
readiness to discharge is brought about, not by a lowering of the thresh¬ 
old—as we are wont, rather inexactly, to say—but by a raising of exci¬ 
taron. This solves a problem that for years was a subject of long discus- 
sions at the institute Erich von Holst and I directed. Then as now we, 
hypothetically, assume that relationships analogous to those which Roe¬ 
der demonstrated as existing on the elementary level—those between 
rest excitability, stimulus effects, and threshold—also prevail on more 
highly integrated levels of the central nervous system, and we symbolize 
them in a thought model (Figure 18b) which does justice to the new dis- 
covery that readiness-increasing stimuli and instantly releasing stimuli 
are not qualitatively different but only quantitatively, that is, in respect 
to the time period between their arrival and their taking effect in the 
efference. This model represents the threshold as being constant: it is 
symbolized in the spiral spring whose tensión is, in this model, not influ- 
enced by an external agent. The opening of the cone valve depends 
exclusively on the pressure of the liquid, in other words, on the present 
level attained by the internal readiness to perform the fixed motor pat¬ 
tern. Immediately releasing stimuli are different from those which 
latently increase readiness only with respect to the time that is necessary 
for them to take effect motorically (Figure 18). A simulation of the real 
process at which the model aims could be made more exact by building 
an element of inertia into the opening and the closing of the valve. 

This is exactly the hypothesis Walter Fíeiligenberg formulated on the 
basis of his experiments with crickets (Gryllus) and fishes (Pelmatochromis 
kribensis) (1974). Fíe says: 

It could be assumed that a certain behavior pattern always occurs when a 
specific concomitant hypothetical physiological State exceeds a certain min- 

3. Analogous Phenomena in Integrated Neural Systems 


imal valué, that is, a critical threshold X 0 , and also that this occurrence 
becomes the more probable the higher the average basic level of this X valué 
is. Spontaneous behavior patterns, like the chirping of a cricket, would be 
characterized, on the basis of this assumption, by the fact that the concomi- 
tant X valué reaches the dimensión of the critical threshold X 0 even without 
the influence of any external stimulus. The occurrence of non-spontaneous 
activities should be dependent on an external stimulation capable of raising 

Figures 18a. and 18b. In the oíd "psycho-hydraulic" model ( left ), the spigot ER 
represents the source of endogenous and automatic generation of stimuli; the line 
Asp symbolizes the present level of action-specific potential. The spiral spring at 
the outlet represents what Roeder called the "stability of the system." The trac- 
tion exerted by the weights SR stands for the effect of the releasing stimuli. In 
Figure 18b, the additional effect of unspecific readiness-releasing stimuli is rep- 
resented by the outflow of the spigots AR. This new model is meant to account 
for the fact that the effect of the specifically releasing key stimulus SR is different 
from that of the endogenous stimulus ER and that of the additional readiness- 
increasing stimuli AR only with regard to its time curve. The difference in the 
height of the two reservoirs containing Asp in the two models is meant to suggest 
that, according to the new hypothesis, the opening of the valve is effected only 
by the necessary pressure from within the reservoir. The simulation could be 
brought closer to the real physiological process by adding a few gadgets, for 
instance, by a mechanism imposing the phenomenon of inertia on the opening 
and the closing of the valve. 


III. The Problem of the "Stimulus" 

their X valué enough to exceed the critical threshold. On this assumption, 
the type of stimulus effect—whether increasing readiness or immediately 
releasing—depends exclusively on the temporal curve of the increment X 
valué caused by it. A strong and rapid increase causes the behavior pattern 
to follow immediately after the stimulus; one should say, in this case, that 
the behavior pattern has been released by the stimulus. A gradual and 
enduring increase would have the effect of allowing later stimulation to 
have an increased chance of lifting the already raised X valué above the 
threshold, thus causing the behavior pattern to be released; in this case one 
would speak of an increased readiness. 

The statements made here by Heiligenberg concern complex behavior 
of intact organisms, a type of behavior which implies, at the very least, 
the function of the complex stimulus-filter of an IRM as well as the func- 
tion of a motor pattern that has a number of intensity-dependent forms 
of appearance. It is interesting to compare Heiligenberg's statements 
with those quoted above from the work of Roeder; all of the latter relate 
to the "behavior" of single neural elements. The similarity of phenom- 
ena occurring at different levels of integration is so striking that one can- 
not help but suspect that at least some of the properties of the integrated 
systems are quite simply due to those of the elements of which they are 
composed. It should also be emphasized that both authors, although 
investigating very different systems, arrive at the inevitable conceptual- 
ization of an "action-specific potential" (ASP). 

Because the modified psycho-hydraulic model shown in Figure 18b 
has, on the continent, been misinterpreted recently in a most amazing 
manner, I feel impelled to prevent a similar misunderstanding among 
readers of the English edition of my book. In the transcript of a discus- 
sion session that took place in my former department at Seewiesen, there 
is a statement to the effect that I myself have revised my oíd model and 
that I now assume that the endogenous "charging up" or accumulation 
of ASP is very small and that, henee, even in the absence of releasing 
key stimuli, no "damming up" of the internal drive need occur. ("Lorenz 
geht davon aus, dafü die endogene Aufladung eines Triebes nur gering ist , das 
hei$t, auch beim Fehlen bedingter Aupernreize kommt es nicht zwangslaufig zum 
Triebstau.") I am thoroughly taken aback that this gross misinterpretation 
went unchallenged and uncontradicted during the discussion itself. 
Anyone with any sense for quantification who looks at the model must 
be able to see that the endogenous production of excitation represented 
as a thick spout is discharging at least ten times the quantity of the three 
dripping spigots of unspecific "charging up" stimuli taken together. 

In consideration of this falsification of my true opinions, it seems 
advisable to emphasize here, once again, what a very crude simplifica- 
tion of the actual physiological process my model—in fact, any model— 
represents. The physiological process could, of course, be simulated 
much more closely by a few additions which have been left out for sim- 

3. Analogous Phenomena in Integrated Neural Systems 


plicity's sake. The most important of these would be a qualification of the 
symbolic valve to indícate an element of inertia, as I have already men- 
tioned above. The valve "sticks" when opening as well as when clos- 
ing—a phenomenon to which Seitz drew attention nearly half a century 
ago. The valve opens very slowly at first, even under strong stimulation, 
and once open it stays open and allows the level of ASP to sink far below 
that level at which it was when the valve first opened. This is precisely 
why instinctive motor patterns tend to "break out" in bouts and not to 
"dribble out" constantly. 

The statements quoted from the writings of Roeder and Heiligenberg 
show that there is one way in which the functions of the integrated Sys¬ 
tems differ sharply from those of their elements, and this is the speed 
with which the changes of readiness caused by stimulation take place. 
Heiligenberg speaks of a strong, but quickly evanescent effect of some 
stimuli as well as of an enduring, slowly fading effect of others. I doubt 
that stimulus effects of similarly long duration can ever be demonstrated 
in isolated neural elements. In complex neural systems, Erich von Holst 
demonstrated that the duration of any change of readiness in a behav- 
ioral system was correlated with its level of integration. Through brain- 
stem stimulation experiments with chickens he found that if a highly 
integrated system had been stimulated, the stimulus-induced increment 
in a readiness for some specific types of activity took a longer time to 
fade away, while the effect of stimulation disappeared almost at once 
when only a subsystem of low integrational level was concerned. If, for 
example, the electrode was positioned so that it activated the entire series 
of motor patterns pertinent to escape from a terrestrial predator, begin- 
ning with a raising of the head, scanning the environment, uttering a 
specific warning cali, and finally taking wing, the bird required several 
minutes to quieten down enough in order to be ready for other types of 
behavior. If, however, the electrode was situated so as to release only 
head-raising and a warning cackle, the aftereffects were of a much 
shorter duration, disappearing quickly after the cessation of the stimulus. 

In intact animáis it can be demonstrated that the "inertia" of integrated 
neural systems is correlated to their complexity, in other words, to the 
number of subsystems between afference and efference. When Seitz 
offered a strongly fight-eliciting dummy to a male Astatotilapia that had 
spent some time in isolation, the fish, in spite of its high internal readi¬ 
ness to fight, took quite some time to run through the sequence of behav¬ 
ior patterns correlated to the rising stages of fighting excitation, as has 
been described in Two/I/3. 

This kind of inertia seems to be due to the number of afferent and 
efferent mechanisms that have to be activated in sequence, one after the 
other, between the arrival of the releasing stimulus configuration and 
the appearance of the activity released. In order to compare the effects of 
readiness-increasing and of directly releasing stimuli, Heiligenberg kept 
a Pelmatochromis kribensis adult male in the company of a number of juve- 


III. The Problem of the "Stimulus' 

niles of the same species. By using a suitable number of fish in a tank of 
a suitable size, it was possible to quantify the average number of attacks 
launched at the juveniles by the adult male in the course of a certain 
amount of time and to gain, in this way, a good measure of the latter's 
readiness to attac_.. When, within this experimental arrangement, the 
adult male was shown a rival or a dummy for a period of thirty seconds, 
no immediate response in the form of typical agonistic behavior was seen 
during this short time, but after the removal of the stimulus and in the 
course of the next three or four minutes, the number of attacks on the 
young fish increased by more than half. This is a typical example of the 
"inertia" of specific excitability. 

When Seitz had enticed one of his Astatotilapia males to fight a rival or 
a rival dummy with great intensity, the sudden removal of the object did 
not abruptly terminate the combative behavior. The fish then regularly 
attacked inadequate substitute objects, such as the small plástic tube join- 
ing the air-disperser to the pressure piping inside its tank. Analogous 
phenomena are observed in many behavior patterns released by compli- 
cated IRMs which obey the law of heterogeneous summation. In some 
cases the inertia of excitation causes displacement activities (the so-called 
"after discharge displacements") to appear after the sudden cessation of 
a stimulus. 

On the other hand, it would seem that behavior patterns released by 
very simple IRMs are better able to respond on very short notice, the 
obvious example being those of escape. In some escape reactions it is pos¬ 
sible to demónstrate, with particular clearness, the difference as well as 
the similarity existing between readiness-increasing and releasing stim- 
uli. When flocks of sparrows (Passer domesticus) or yellowhammers 
(Emberiza citrinella) accumulate on an open field or road, having been 
lured far away from any cover by a supply of food, they appear to be 
struck by blind panic at regular intervals, flying back as fast as they can 
to the cover, only to return again to their feeding place almost at once. 
The stimulus situation presented by exposure to flying predators, partic- 
ularly the sparrow hawk (Accipiter nisus), is definitely not directly releas¬ 
ing these escape reactions, although it does exact extreme watchfulness. 
In other words, being so exposed increases readiness to an extreme, and 
to such an extent that the escape threshold is reached again and again, at 
short intervals, even though the typical releasing stimulus, represented 
by a fast-moving object silhouetted against the sky, never appears. This 
also explains (as was mentioned in Two/I/7) why it is impossible to mea¬ 
sure a constant threshold of escape reactions. 

4. Action-Specific Potential (ASP) 

Erich von Holst discovered and analyzed the effects of the spontaneous 
generation of impulses and their coordination within the central nervous 

4. Action-Specific Potential (ASP) 


system of spinal fishes. He also studied another important phenomenon 
in the spinal sea horse (Hippocampus), a phenomenon which had previ- 
ously been described by Sherrington and termed by him the "spinal con- 
trast." A spinal sea horse kept alive by artificial respiration shows no 
movements of the dorsal fin, the main locomotor organ of this fish. The 
fin remains quite still, but not in the position it assumes in an intact sea 
horse during repose, when the fin rests completely folded at the bottom 
of a groo ve along the fish's back; in the spinal preparation the dorsal fin 
remains partially unfolded. By applying certain stimuli, simply by exert- 
ing, for example, pressure on the "neck" región of the fish, it is possible 
to cause the dorsal fin to fold down completely into its groove, just as it 
does in the intact organism when at rest. When the stimulus is stopped, 
the fin not only unfolds, but extends further than it had before the stim- 
ulation was applied. The longer the stimulus causing the fin to fold is 
applied, the higher it rises after the stimulus has been removed. If, 
through a sustained application of the stimulus, the fin is forced into its 
folded position for a much longer period of time, it will not only unfold 
completely when the stimulus is removed, but it will also—if only for a 
very short while—begin to perform the undulating movements of loco- 
motion. After ceasing to undulate, it sinks down gradually in an asymp- 
totic curve to the "half-mast" position it usually assumes in the spinal 

These phenomena von Holst interprets hypothetically as follows: 
While the centrally coordinated motor pattern of locomotion is at rest, 
an action-specific exciting agent—perhaps in the form of a specific neu- 
rohormone—is being continuously produced, and a performance of the 
motor pattern consumes part of this accumulation. The quantity of con- 
sumption is dependent on the intensity of performance; even the slight- 
est intention movement uses up a little of it. The quantity of endogenous 
production of excitability specific to a certain motor pattern is correlated 
to the average rate at which it is used by the organism. In a wrasse 
(Labridae), the endogenous production of impulses keeps the fish swim- 
ming more or less continuously during the hours of daylight; in a sea¬ 
horse, which normally swims but a few minutes each day, a much 
smaller production is sufficient. As long as the inhibitory mechanisms at 
work in an intact seahorse keep the dorsal fin completely folded in its 
groove, all of the centrally produced excitability is accumulated so that, 
on the sudden removal of the inhibition, enough of it is available to per- 
mit the fish to swim away. In the spinal preparation, the central inhibi¬ 
tion is lacking and the endogenous production of impulses finds its way 
to the motor cells unimpeded, causing a movement of low intensity that 
takes the form of a partial raising of the fin; this consumes exactly as 
much excitation as is continuously furnished by the endogenous impulse 
production. In order to accumulate the amount of excitability to the 
threshold level at which the undulating swimming movements set in, it 
is necessary to prevent the excitability from "seeping out" through the 


III. The Problem of the "Stimulus' 

pathway of a low-intensity movement. This is achieved by substituting 
an artificial stimulus for the central inhibition that is lacking in the 
spinal preparation. 

These functions of the spinal cord of a fish, described by Erich von 
Holst, are strictly analogous to those processes observed in the behavior 
of intact animáis that were discussed in Two/I/ 6, 7. 

I do not see how the assumption can be avoided that all these phenom- 
ena are caused by the spontaneous and continuous production, during 
the quiescence of a motor pattern, of "something"—whatever it may 
be—that is consumed or otherwise eliminated by a performance of the 
pattern. This hypothesis is strongly supported by the fact that we find 
exactly the same changes in readiness, the same relations between inter¬ 
nal readiness and external stimulation, in some other activities which 
depend on the accumulation of "something" that we know and that we 
can measure quite well. Examples of such activities are those regulating 
the contents of hollow organs, as well as those controlling tissue needs. 

The motor patterns of urination performed by a male dog show all the 
phenomena here under discussion. A very strong releasing stimulus sit- 
uation, such as the smell of a rivaTs mark in the dog's own territory, will 
cause him to lift his leg even when the amount of uriñe at his disposal 
is, at the moment, negligible. Even under the pressure of a much higher 
urinating potential, the dog will still look for releasing stimulus situa- 
tions, such as upright objects, preferably on exposed corners, at which to 
lift his leg. Under extreme internal pressure he will forgo every external 
stimulation and even forget the conditioned inhibition of house training 
and urinate on the carpet—in this pitiable situation usually without even 
lifting his leg. Adherents of stimulus-response psychology have con¬ 
tended that sexual behavior in male mammals and in man is similarly 
dependent of the pressure within the seminal vesicle and, on the basis 
of this assumption, have coined the term "detumescence drive." As we 
have just seen, this assumption is not entirely correct, even with regard 
to the urination activities of the male dog. 

With motor patterns that supply tissue needs, such as eating, drinking 
and, quite particularly, breathing, a déficit in the concentration valué of 
the required Chemical within the tissues acts analogously to the pressure 
within hollow organs or, for that matter, to the level of the hypothetical 
liquid accumulated in the tank of our thought model. But things are 
more complicated than this because, besides the signáis emitted by the 
tissues needing supply, the endogenously produced readiness to perform 
each of the motor patterns concerned is adding its own contribution to 
the organism's general State of readiness—as de Ruiter (1963) and 
Dethier and Bodenstein (1958) have demonstrated. Moreover, the pro- 
prioception of the contents of hollow organs may also contribute. None 
the less, it is a perfectly justifiable assertion to say that the intensity of 
appetitive behavior, as well as that of the consummatory acts released. 

4. Action-Specific Potential (ASP) 


rises relative to the extent to which the Chemical in question is deficient; 
to a comparable extent, the selectivity of the IRMs concerned decreases. 

Breathing also serves to supply tissue needs, and since the processes 
regulating it are simpler than those governing eating and drinking, the 
analogy to our thought model, to which Haldane (1932) has drawn atten- 
tion, is even more striking. Carbón dioxide is continuously produced in 
the body and stimulates the breathing center whenever its concentration 
rises above a certain threshold valué. The motor pattern then released 
eliminates the carbón dioxide. 

We do not yet know of what that problematical "something" consists, 
that "something" which is produced while a motor pattern is at rest and 
which is consumed when it is in action, but we do know that it does exist 
and we know that it is specific, in every single instance that has been 
investigated, to one particular motor pattern—in other words, specific to 
one of those sequences of movements which is species-characteristic and 
which, as Whitman and Heinroth discovered, can be homologized just as 
definitively as morphological characters. We know that the eífects of 
external stimulation and of the endogenous build-up of excitability com¬ 
bine in a manner that produces qualitatively identical eífects indepen- 
dently of the proportion each of these apparently so diíferent factors has 
contributed to the total sum: strong internal readiness and weak external 
stimulation add up to exactly the same result as a very strong stimulation 
impinging at a time when excitability is at a low ebb. The quality of the 
motor pattern released remains the same independently of the valué 
attained by this addition; it is only its intensity that fluctuates in propor¬ 
tion to this valué, ranging from hardly perceptible intention movements 
to the full teleonomic performance of the motor pattern. As has been 
explained in Two/I/2-6, the prerequisite for discovering these demon- 
strable relationships was knowing about all the stimulus configurations 
acting as key stimuli on the same IRM and, furthermore, being com- 
pletely familiar with all the forms which a motor pattern could assume 
at diíferent intensities. Stimulus configurations that are known to act as 
key stimuli on one IRM and to exert one qualitatively identical and quan- 
tifiable influence, add their effects to those of an equally specific and 
qualitatively identical and quantifiable State of endogenous excitability; 
the sum of this addition is indicated by the intensity of the motor pattern 
performed, which varies only unidimensionally along a single scale of 
intensities, which again permits perfect quantification: thus can Seitz's 
law of heterogeneous summation be restated. 

In the terms of our hydraulic model, the quantity of discharge at the 
outflow is determined by the sum of endogenous and exogenous factors, 
while its quality is constant and quite independent of the proportions in 
which they contribute to the sum. This constancy of specific quality is 
remarkable in view of the heterogeneity of the influences contributing: 
no matter how they are mixed, they find their exclusively quantitative 


III. The Problem of the "Stimulus' 

expression in the unidimensional variation of the action's intensity. This 
unidimensionality implies the existence of a firmly integrated system 
that ensures the discharge of highly differentiated motor patterns in the 
exact environmental situation to which they are adapted. The function 
of this system does indeed fully merit Portielje's oíd term, "action-and- 

The heterogeneity of afferent influences on the one hand and, on the 
other, the unity of internal readiness as well as the unity of the subse- 
quent motor discharge, appear to imply that the specificity of the whole 
process is dependent less on the specificity of the contributing factors 
than on the "reservoir" in which they are "collected." We do not know 
what, physiologically, represents that reservoir, ñor, indeed, what it is 
that is collected in it. But as has already been pointed out, we find the 
most detailed analogy of function with those of processes regulating the 
contents of hollow organs or of supplying tissue needs. However one 
tries to construct a flow diagram or a thought model simulating the pro¬ 
cesses just discussed, one will find it impossible to avoid the assumption 
of an accumulating action-specific "readiness" or excitability. The 
accepted expression, "action-specific potential," (ASP) seems to me to be 
a neutral and useful term. The only danger of misunderstanding consists 
of a possible confusión of the ASP of a fixed motor pattern with the mem- 
brane potential of a single neural element. Great as the analogies are, it 
is not permissible to equate the two concepts. 

Chapter IV 

The Behavior Mechanisms Already 
Described Built into Complex 

1. Appetitive Behavior Directed at Quiescence 

As has been explained in the Introductory History and in One/I/1, and 
mentioned again in Two/II/1, both Oskar Heinroth and I regarded the 
arteigene Triebhandlung, the species-characteristic drive action, as the 
smallest and ultímate, indivisible unit of behavior. But this unit is 
actually a quite complex mechanism comprising appetitive behavior, the 
function of an IRM, and the performance of a consummatory action. The 
conceptualizations of appetitive behavior, the innate releasing mecha¬ 
nism, and the fixed motor pattern have been proved applicable in many 
cases in which these functions are built into sequences of a kind that are 
very different from, and more complicated than, the species-characteris¬ 
tic drive action. If Heinroth and I regarded the tripartí te sequence of 
appetitive behavior, IRM, and the consummatory act as a single, and the 
only element of animal behavior, this must be regarded as a simplifica- 
tion that is not only forgivable, within a new and developing branch of 
Science, but also unavoidable and heuristically fertile. For, after all, when 
drawing flow diagrams, biocyberneticists intentionally rely on similar 
simplifications. Flow diagrams representing these kinds of complicated 
behavior systems, that is, appetitive behavior, IRMs, and consummatory 
acts, are so similar to each other that it is permissible to speculate that 
the mechanisms underlying their functions are not only analogous but 
physiologically akin to one another. 

In his classic treatise on appetites and aversions as constituents of 
instinct, Wallace Craig (1918) was the first to demónstrate that the very 
same elements, such as goal-directed appetitive behavior, innate releas- 


IV. Complex Systems of Behavior Mechanisms 

ing mechanisms, and fixed motor patterns can be joined together in quite 
different ways to form teleonomic systems of behavior. He defines appe- 
tites and aversions in the following manner: an appetite is a State of 
arousal which continúes as long as a specific stimulus situation, which 
he calis an "appeted" stimulus, is not reached. When this specific stim¬ 
ulus situation is reached, however, the consummatory action is set free, 
the appetite is assuaged, and a State of satiety, that is, of relative quies- 
cence is achieved. 

Aversión, on the other hand, has been defined by Craig as a State of 
agitation which continúes as long as a certain specific stimulus, referred 
to as the disturbing stimulus, is present, but which ceases and is replaced 
by a State of relative rest when the stimulus has ceased to act on the 
animal's sense organs. As an example of aversión, Craig cites: 

. . . the so-called jealousy of the male dove, which is manifested especially in 
the early days of the brood cycle before the eggs are laid. At this time, the 
male has an aversión to seeing his mate in the proximity to another dove. 
The sight of another dove near his mate is an "original annoyer" (Thorn- 
dike, Chapter IX) [1911]. If the male sees another dove near his mate, he 
follows either of two courses of action: namely a) attacking the intruder with 
real pugnacity; b) driving his mate, gently, not pugnaciously, away from the 
intruder. When he has succeeded either in conquering the stranger and get- 
ting rid of him, or driving his mate away from the stranger, so that he has 
got rid of the disturbing sight of another dove in the presence of his mate, 
his agitation ceases. If we prevent him from being successful with either of 
these methods, as by confining the pair of doves in one cage and the third 
dove in plain sight in a contiguous cage, then he will continué indefinitely 
to try both methods. Or if we leave all three doves free in one pen, the mated 
male will try the mettle of the intruder and conquer him if he can; if he 
fails, he will turn all his energies to drive his mate away from the intruder. 

Or if in former experiences he has learned to gage this individual intruder, 
if he has conquered him before, he will promptly attack him now, but if 
defeated by him before, he will now choose the alternative of driving away 
his mate. In sum, the instinctive aversión impels the dove to thoroughly 
intelligent efforts to get rid of a disturbing situation." 

An essential difference between appetites and aversions exists with 
regard to the reinforcing and to the extinguishing function of releasing 
stimuli. In the case of appetites, the finding of any slight stimulus which 
adds to the stimulus situation, releasing the consummatory action, has 
the effect of a reinforcement. Any addition to the stimulus situation that 
helps to increase the readiness for a specific act, as discussed in Two/III/ 
4, will "encourage" the animal in its efforts. Conversely, in aversions, 
every diminution of the disturbing stimulus situation will reinforce the 
kind of behavior that has led to this relief. 

The concept of aversión, as defined through these points of view, does 
not seem very satisfactory when applied to a certain, rather common type 
of behavior which may be classed among the aversions because it serves 

2. Searching Automatism 


to make the organism avoid certain detrimental stimulus situations. All 
the physiologic mechanisms that cause the animal to choose the sort of 
habitat to which its species is adapted, in other words, that cause it to 
search for a special optimum of humidity, temperature, or illumination 
and the like, obviously are in complete accordance with Wallace Craig's 
definition of aversions. There are, however, very many other instances 
in which appetitive behavior is striving for a very special stimulus situ- 
ation that, far from releasing any consummatory action, is just what the 
animal requires in order to come to rest. Thus, the mere absence of such 
a stimulus situation acts as an "original annoyer," but no gradation of 
stimulus-strength leads the organism along the correct path for getting 
rid of it. What the animal does, in this case, is the purest form of blind 
searching. An example is afforded by the behavior of many birds dwell- 
ing in dense masses of reeds, such as the bearded tit (Panurus biarmicus) 
or the least bittern (Ixobrychus minutus). These birds are content only in 
an environment which permits them to sit within dense cover, both feet 
turned outwards with claws clasping, on either side, the vertical stems of 
sedge. A fully fledged young greylag goose which has strayed from its 
family will most strenuously search for it, wandering about restlessly, 
sounding distress calis, and never coming to rest until it has found its 
loved ones again. 

To describe this kind of appetitive behavior as aversión seems some- 
what strained: to say, for instance, that the bittern has an aversión to all 
landscapes that are not a dense growth or mass of reeds, or that the 
young goose has an aversión to all environments not containing its par- 
ents. Monika Meyer-Holzapfel has proposed describing this kind of 
behavior as appetite directed at quiescence (1956). Physiologically it is 
different from the true aversión in that there is no gradient leading the 
animal away from the disturbing stimulus and towards the situation in 
which it can find rest. The term "quiescence" must never be misinter- 
preted to mean that the animal, having got rid of the disturbance, will 
now lapse into a State of complete rest or fall asleep. On the contrary, 
being liberated from the disturbance, it will proceed to perform any of 
a great number of behavior patterns which, up to that moment, have 
been suppressed by the disturbing stimulus. 

Transitions and intermedíate stages between aversions and appetites 
for quiescence are, of course, numerous. When we go indoors on a coid 
winter's day, it remains questionable whether we do so because we feel 
an aversión to the coid, or whether we are moved by the longing for a 
warm stove. 

2. Searching Automatism 

Monika Meyer-Holzapfel's conceptualization of appetitive behavior as a 
striving for quiescence (1956) called attention to the fact that Heinroth's 


IV. Complex Systems of Behavior Mechanisms 

species-characteristic drive action (Two/I/1), that is, a system composed 
of appetitive behavior, an IRM, and a consummatory act did not, by any 
means, represent the one and only way in which these three elements 
could be combined. 

Prechtl and Schleidt (1950, 1951) demonstrated a very special system 
composed of these three parts. As long as a newly born kitten is awake 
and not sucking at one of its mother's teats, it performs a constant, to- 
and-fro sideways sweeping movement with its head and the foreparts of 
its body. All the while it is doing this it is also creeping slowly forward 
and, when it touches a solid vertical surface, will snuggle up to this and 
maintain contact with this while continuing on its way. If the kitten 
comes into contact with fur, it will stop its forward motion and orient 
the sweeping movements so that its nose remains pressed against the fur 
while moving to and fro in it. When by doing this the kitten happens to 
touch a spot bare of any fur, the sweeping movements cease and are 
replaced by repeated snapping movements of the mouth, which under 
normal conditions will find the teat. When this is accomplished, all 
movements hitherto described come to a standstill and, while the upper 
jaw and the tongue endose the nipple, the kitten now begins to suck and 
to perform a new motor pattern, thrusting its nose rhythmically against 
the mother's breast and massaging the breast gently with its fore paws— 
a movement termed "milk treading" that is found in many mammals, 
including man. 

The beginning of the chain of activities is, in this instance, the unin- 
hibited performance of a fixed motor pattern that continúes un til a specific 
stimulus situation is reached which inhibits it while, at the same time, 
others are released. In this chain of processes, fixed motor patterns play 
the role of appetitive behavior insofar as the stimulus situation which 
terminates them at the same time serves to release another fixed motor 
pattern which forms the next link of a functionally consistent chain. 

H. Prechtl (1955) has shown that the activities of searching for the 
mother's breast performed by the human baby, and the reaction at find- 
ing the nipple, are programmed in a very similar way, except for an 
additional orienting mechanism: if one gingerly touches the cheek of the 
baby while it is moving its head to and fro in the automatic movement, 
it interrupts this rhythm suddenly to perform a well-oriented snap in the 
direction from which it has been touched. 

The program of these activities accomplishes a highly teleonomic 
sequence in which fixed motor patterns and innate releasing mecha¬ 
nisms are linked together to form an adaptable program. The sequence 
begins, in this instance, with the performance of a fixed motor pattern 
which is not under any constant inhibition and does not stand in need 
of an IRM to be liberated. One motor pattern in this chain of activities 
plays the role of an appetitive behavior striving for the next motor pat¬ 
tern: the stimulus situation achieved not only terminates it, but, at the 

3. Hierarchical Systems 


same time, corresponds to the IRM setting off the next action of the 

3. Hierarchical Systems 

Chains of behavior mechanisms such as the one analyzed by Prechtl and 
Schleidt (1950, 1951) are extremely common; in fact, they are much more 
common than actions consisting of only a single link of appetitive behav¬ 
ior, one IRM, and one comsummatory action. It is perfectly legitímate— 
it even constitutes a mark of genius—to discover the simplest possible 
instance in which a natural law is realized, as, for example, are the Men- 
delian laws in monogene hybrids. Yet one must not commit the error of 
believing that the simplest instance is also the most frequent one—as 
Oskar Heinroth and I admittedly did in the case of the species-character- 
istic drive action. Further analysis has shown that much more complex 
systems are much more common; at the same time, this analysis has dem- 
onstrated the valué of our first conceptualizations: all these complex Sys¬ 
tems have been proved to be based on the very same three elementary 
mechanisms that we assumed in our provisional flow diagrams. 

Only rarely does the primary appetitive behavior, with which an ani¬ 
mal begins an activity, lead directly to a stimulus situation releasing the 
consummatory act terminating that activity. Much more commonly the 
primary appetitive behavior leads to a situation in which the first form 
of appetitive behavior is switched off and a subsequent one is switched 
on, just as in the case of the kitten finding the nipple area. The program 
of such a chain is much more adaptable than the single tripartite action, 
not only because the time spent performing each phase is variable, but 
also because every step of appetitive behavior usually ineludes orienta- 
tion or other mechanisms exploiting instant information—as will be dis- 
cussed in Chapter VI. 

Tinbergen describes this type of chain as the hierarchical organizaron of 
instinct. One simple example is found in the behavior of the hobby fal- 
con (Falco subbuteo), which generally direets its hunting flight to areas 
promising prey. On discovering a flock of potential prey—for instance, 
a flock of starlings—the falcon approaches and thereby causes the star- 
lings to cluster cióse together. This formation releases in the falcon a 
"sham attack," that is, a motor pattern which is not adapted to catching 
prey but is adapted to stampeding one starling out of the swarm. If this 
is successful, the new situation releases true prey catching, a motor pat¬ 
tern calculated to hit the prey with great impact, usually from below. 
This action bears the character of a consummatory act, indeed much more 
so than the subsequent consummation, that is, eating the prey. After 
striking and making the kill, the falcon stands on the dead bird, often 
for quite a while, before beginning to pluck; afterwards, the eating it 


IV. Complex Systems of Behavior Mechanisms 

does with unhurried movements provides the impression that these dis- 
passionate actions do not really belong to the same behavioral system as 
those of the catching and killing. 

Tinbergen has represented the principie of such a hierarchical orga¬ 
nizaron in a diagram which intentionally simplifies matters but, never- 
theless, very ably demonstrates the interactions conceptualizing the 
duality of the endogenous generation of excitability on one side, and of 
the exteroceptor function of the IRMs on the other. He uses the concept 
of a "central excitatory mechanism" (CEM), as proposed by Frank Beach 
(1942), though he specifically opposes the notion of monocausality, that 
is, of regarding the CEM as the single, or even as the main source of 
stimulation. Concerning this, Tinbergen (1951) says: 

. . . it seems that the single CEM postulated by Beach is rather a system of 
CEMs of different levels. Each 'centre' in our system is a CEM in Beach's 
sense, as each of these centres has its own afferent and efferent connexions. 

Of no less importance is the difference between motivational and releas¬ 
ing factors. For, as we have seen, the motivational factors influence the CEM 
itself while the releasing factors activate a reflex-like mechanism, the IRM, 
removing a block that prevented the outflow of impulses along the efferent 

The system C(entral) E(xcitatory) M(echanism)-I(nnate) R(eleasing) 
M(echanism) is tentatively presented in Figs. [19a and 19b]. Let us first con- 
sider Fig. [19a], which represents one centre of an intermediate level. 

The centre is 'loaded' by motivational impulses of various kinds. First it 
receives impulses from the superordinated centre of the next higher level. 
Impulses from this higher level flow into other centres as well, in fact to all 
the centres controlled by the higher centre. Second, centre 1 may receive 
impulses from an 'automatic', self-generating centre belonging especially to 
it ([compare] . . . the dual nature of centres as found by von Holst. . .). Third, 
a hormone might contribute to the motivation, either by acting directly on 
centre 1, or through the automatic centre. . . . Fourth, internal sensory stim- 
uli . . . may help to load centre 1. Fifth, external sensory stimuli might also 
act directly upon the centre and contribute to its motivation. 

This system together represents a CEM in Beach's sense, belonging to one 
level of the hierarchical system. 

Outgoing impulses are blocked as long as the IRM is not stimulated. When 
the adequate sign stimuli impinge upon the reflex-like IRM, the block is 
removed. The impulses can now flow along a number of paths. All but one 
lead to subordínate centres of the next lower level. However, all these 
centres are prevented from action by their own blocks, and most of the 
impulses therefore flow to the nervous structures controlling appetitive 
behaviour. This appetitive behaviour, as we have seen . . ., is carried on until 
one of the IRMs of the lower level removes a block, as a result of which free 
passage is given through the corresponding centre of this next lower level. 
This 'drains away' the impulses from the appetitive behaviour mechanism 
and conducís them to the appetitive behaviour mechanism of the lower 

3. Hierarchical Systems 


Centres of next 
lower leve! 

Externa) \ 
motívatíonal j 
impulses J 
Intemal stimuli 
imputees from \ 
same leve! 

Intrinsíc motívational 
impulses from superordinated 

Block preventing 
continuous discharge 

Appetitive behaviour 
pattern controlled 
by centre 1 

Figure 19a. Tentative representation of an instinctive 'centre' of an intermedíate 
level. Explanation in the text. (Tinbergen, N.: The Study of Instinct.) 

Fig. [19b] suggests how centres of this type might be organized within 
one major instinct. The reproductive instinct of the male three-spined stic- 
kleback has been taken as a concrete example. The hormonal influence, pre- 
sumably exerted by testosterone, is acting upon the highest centre. This 
centre is most probably also influenced by a rise in temperature. These two 
influences together cause the fish to migrate from the sea (or from deep fresh 
water) into more shallow fresh water. This highest centre, which might be 
called the migration centre, seems to have no block. A certain degree of 
motivation results in migratory behaviour, without release by any special set 
of sign stimuli, which is true appetitive behaviour. This appetitive behav¬ 
iour is carried on—the fish migrates—until the sign stimuli, provided by a 
suitable territory (shallow, warm water and suitable vegetation) act upon the 
IRM blocking the reproductive centre sensu stricto, which might be called 
'territorial centre'. The impulses then flow through this centre. Here, again, 
the paths to the subordinated centres (fighting, nest-building, etc.) are 
blocked as long as the sign stimuli adequate to these lower levels are not 
forthcoming. The only open path is that to the appetitive behavior, which 
consists of swimming around, waiting for either another male to be fought 
or a female to be courted, or nest material to be used in building. 

If, for instance, fighting is released by the trespassing of a male into the 
territory, the male swims towards the opponent (appetitive behaviour). The 
opponent must give new, more specific sign stimuli, which will remove the 
block belonging to one of the consummatory acts (biting, chasing, threat- 
ening, etc.) in order to direct the impulse flow to the centre of one of these 
consummatory acts. 


IV. Complex Systems of Behavior Mechanisms 

Figure 19b. Hierarchical centers of the major reproductive instinct of the stickle- 
back male. Motivational impulses are represented by straight arrows which 'load' 
the centers (shown here as circles). These impulses may come from the external 
environment as well as from superordinated centers, or they may occur sponta- 
neously within a center (this is not considered in the schema). The shaded rec- 
tangles indicate inhibiting influences, which prevent a continuous discharge of 
motor impulses. These blocks are removed by innate releasing machanisms. 
When this has occurred the animal will show a specific appetitive behavior until 
more specific releasing stimuli activate the next subordinated instinct and the still 
more specific appetitive behavior. The concave, two-headed arrows between cen¬ 
ters of the same level indicate inhibiting relationships and the existence of DIS- 
PLACEMENT ACTIVITIES. Below the level of the consummatory acts a number 
of centers come into action simultaneously. The relation between subconsum- 
matory centers of the same level is indicated by horizontal lines. (Additional 
explanation in the text.) (Eibl-Eibesfeldt, I.: Ethology: The Biology of Behavior.) 

3. Hierarchical Systems 


The various centres located at the various levels are, therefore, not orga- 
nized in exactly the same way. The very highest centre has no block. If there 
were blocks at these very highest centres, the animal would have no means 
of 'getting rid' of impulses at all, which, as far as we know, would lead to 
neurosis. The next centre responds, in comparison to the lower centres, to 
a relatively higher number of motivational factors. 

The next centres of the male stickleback's reproductive behaviour pattern, 
represented, for example, by fighting and nest-building, are loaded primar- 
ily by the impulses coming from the higher centre. Whether there are spe- 
cial motivational factors for each of these centres besides those coming from 
the higher centre is not certain, but I think there are, because the fighting 
drive displays the phenomenon of after-discharge, and, moreover, seems to 
be motivated (not merely released) by external stimuli. In general it seems, 
that the lower we go, the more pronounced the influence of external releas¬ 
ing stimuli becomes. 

Arrows between the centres of one level indicate interrelationships sug- 
gested by the existence of mutual inhibition and of displacement activities. 

It should be emphasized that it is quite possible that these interconnexions 
do not in reality run directly from one centre to the other, but go by way of 
the superordinated centre, and that the 'inhibition' of one centre by the 
other may in reality be competition, the 'inhibiting' centre by decrease of 
resistance 'draining away' the impulse flow at the expense of the 'inhibited' 

Thus the motivation is carried on through a number of steps, which may 
be different in different instincts, down to the level of the consummatory 
action. Here the picture is changed. When the block of the consummatory 
centre is removed, a number of centres come into action simultaneously, 
between which horizontal co-ordinative forces are effective. With fish, these 
centres below the consummatory level are arranged in two or three planes, 
the lowest of which is the centre of the right or left fin-ray muscle, which 
is a relatively simple type of nervous centre. The relation between sub-con- 
summatory centres of the same level are represented by horizontal lines. 

Tinbergen stresses that, in his opinión, "these diagrams represent no 
more than a working hypothesis of a type that helps to put our thoughts 
in order"; also, he emphasizes that the diagram does not take into 
account the great number of feedback effects and regulating cycles con- 
necting the lowest and the highest levels of any hierarchical organiza- 
don of instinct. It is certain, for instance, that a high ASP of a motor 
pattern at the lowest level can exert a decisive influence on the highest 
"centers." These qualifications notwithstanding, it is necessary to State 
that Tinbergen's diagram is indeed much more than a mere working 
hypothesis. The stratification of superimposed levels of integration, as 
well as the IRMs shunting behavior from one to the other, are realities. 
Their number can and has been ascertained and the specific stimulus sit- 
uations causing the switch from one level of appetence to the subsequent 
one have been experimentally analyzed. 


IV. Complex Systems of Behavior Mechanisms 

Higher level ¡nstinct 

Lower level ¡nstinct 

Fixed action pattern 

Muscle movement 

Figure 20. The hierarchical scheme of G. P. Baerends (1956), clearly showing the 
interrelationships of the centers. Especially noteworthy is the fact that lower cen- 
ters are often controlled by several higher ones. Dotted lines represent inhibitory 
relationships between mechanisms of the same order. (Eibl-Eibesfeldt, I.: Ethology: 
The Biology of Behavior.) 

In his analysis of the parental care exhibited by the female diggerwasp 
(Ammophila campestris) (1941), Baerends found that subsystems of the 
hierarchy, which are situated on a lower level, are quite often governed 
by more than one locus belonging to a higher level. It is hardly worth 
mentioning that this is frequently the case on levels below that of the 
fixed motor pattern, since, obviously, the contraction of a certain muscle 
can occur in the Service of many functions. An analogous relationship on 
the level of motor patterns appears more interesting. As has already been 
mentioned in Two/I/12, there are a number of fixed motor patterns 
which can be performed in the context of many different systems of 
behavior. These "multi-purpose activities" are, for the greater part, rather 
simple tools such as those of locomotion, et cetera. Baerends discovered 
highly differentiated motor patterns, each of which was adapted to a spe- 
cial function, but, in this function they recurred again and again in dif¬ 
ferent parts of the temporal sequence determined by the hierarchical 
organization. Baerends's diagram, represented in Figure 20, takes into 
account all the pertinent aspects observed in the behavior of the female 
diggerwasp (Ammophila campestris Jur.). Baerends writes: 

During the first bright days in June the males leave the cocoons. The females 
appear some days later. Mating takes place within a few hours after the 
female has hatched. Shortly after coition the female begins to dig her first 
nest (nest A). As nesting sites fíat, sandy areas with a compact soil are 
selected. The nest consists of a vertical shaft about 2-1/2 cm long and one 
elliptical cell about 2 cm in length. Having finished the nest, the shaft is 

3. Hierarchical Systems 


temporarily closed by loosely filling it up with some clods, and the wasp 
then disappears into the heather. She usually returns within a few hours, 
carrying a paralyzed Caterpillar. She reopens the nest, carries the Caterpillar 
down and lays an egg. Then she closes the nest with much more care than 
before. After she has finished, it is impossible for the human eye to distin¬ 
guiste the entrance from the surroundings. She now leaves nest A and it may 
be some days, before she provisions it again. 

Soon afterwards the wasp begins to dig a new nest (B). 

It appeared to be a general rule that once a nest was begun, the work at 
this nest was continued without interruption, until an egg was laid. 

The first series of activities, therefore, constituting the first stage of the 
care for every nest made, will be called the first phase. 

After completing the first phase of nest B, which takes her one or more 
days, dependent on the amount of sunshine, the wasp returns to nest A and, 
before fetching a Caterpillar, opens the nest. After a brief visit, she closes it 
again and flies off to the heather. Such a visit, in which no Caterpillar is 
brought, will be called a "solitary visit", in contrast to "provisioning visit". 
As a rule the wasp, after this solitary visit brings one or more fresh caterpil- 
lars, before she leaves the nest alone for the second time. This phase I shall 
cali, therefore, the second phase; it consists of a solitary visit followed by 
storing 1-3 caterpillars. Occasionally this phase consists of a solitary visit 
only, namely when the egg has not yet hatched at the time of this visit. 

After having finished this second phase in nest A, the wasp carries 
through the same phase in nest B. Now she again pays a solitary visit to nest 
A and there enters into the third and last phase, consisting of one or more 
solitary visits and the storeage [sic] of another 3-7 caterpillars. This phase is 
concluded by closing the nest in an especially careful manner, the wasp 
pressing down the contents of the shaft with her head, during which a loud 
humming sound can be heard. 

She now goes to nest B to accomplish the third phase here. After having 
finally closed this nest she begins to dig a new nest. 

As we see from the above, in each nest provisioning occurs in three 
phases. During every phase the wasp is occupied with one particular nest 
exclusively, interrupting work at the nest only for foraging on her own 
behalf, for hunting caterpillars or for sleeping, but never for any work con- 
cerning another nest. 

Having finished one phase she goes to another nest and works through 
an entire phase there. If there is no other nest to provide for, she makes a 
new nest. Occasionally, in very favourable weather, the wasp, after having 
completed the second phase in nest A, digs a new nest C before beginning 
the second phase of nest B. In this way a wasp sometimes has three nests 
under her care. 

The second and the third phase always begin with a solitary visit; some¬ 
times a phase consists of even no more than that one solitary visit. This may 
occur when the nest has been disturbed shortly before this visit, or, in the 
case of the second phase, when the egg has not yet hatched. This suggests 
that the solitary visit serves as an inspection, that is to say that the wasp 
during this visit receives stimuli from the contents of the cell which deter¬ 
mine whether she will leave the nest alone or go and fetch fresh caterpillars. 


IV. Complex Systems of Behavior Mechanisms 

An experimental test of this hypothesis is possible on the following basis. 
If the solitary visit actually has a regulating function, it should be possible 
to influence the wasp's subsequent behavior by changing the contents of the 
nest just before the solitary visit. This could not be done in the real nests but 
it appeared that the wasps did not interrupt their provisioning activities, 
when I replaced their nests by artificial nests, provided certain precautions 
were taken. These nests were made of gypsum and consisted of a lower part, 
containing the cell, and a lid that could easily be lifted so that I could reach 
the cell and change its contents at will. 

I carried out the following experiments: 

1. Nests which according to preceding observations, should be provisioned 
immediately after the solitary visit, were disturbed by removing the larva 
just before that visit. The result was that the nest was abandoned after the 
first solitary visit. 

2. In similar nests, I replaced the larva by a paralysed [sic] Caterpillar with 
an Ammophila's [sic] egg, taken from another nest. Now the wasp did not 
start provisioning immediately after the solitary visit (as she should have 
done), but she waited until the egg was hatched. 

3. Before the wasp paid her first solitary visit of the third phase to the nest 
I added some paralysed caterpillars to the contents of the nest. The result 
was that the wasp either stopped provisioning altogether, or at least 
brought less caterpillars than the smallest amount ever stored under nor¬ 
mal conditions. 

4. In nests containing one Caterpillar with an Ammophila's egg, nests, there- 
fore, that should not be provisioned immediately, I replaced the egg by 
a larva. In these cases the wasp brought fresh caterpillars soon after the 
solitary visit. 

5. Occasionally a wasp pays a solitary visit when the third phase is halfway 
concluded. A few times I succeeded in taking all caterpillars away just 
before the visit. Normally the wasps should have brought only a few 
more caterpillars, but now they again stored a considerable number of 
caterpillars, making the total amount of stored food larger than ever 
observed under normal conditions. 

The same experiments were carried out just before provisioning visits. 
Under these circumstances the wasps did not react to any change in the con¬ 
tents of the cell. They even continued provisioning when the larva was 
removed together with the food. 

These experiments conclusively show, therefore, that the solitary visit is 
a real inspection, during which the wasp learns how to act in the following 
hours or even days. 

Apart from having a function as a regulatory principie, the solitary visit 
also demonstrates a most remarkable phychological fact: although the exter- 
nal stimulating situation is exactly the same at a solitary visit as at a provi¬ 
sioning visit, the wasp's behaviour is profoundly influenced by it at the first 
occasion, whereas at the second occasion not the slightest influence can be 

3. Hierarchical Systems 


In one case, however, the contents of the nest does influence the wasp's 
behaviour during a provisioning visit. When I put certain objects into the 
cell just before the wasp would pay her very first visit, during which she 
had to lay her egg, the wasp did react. If the object was a Caterpillar with an 
Ammophila's [sic] egg or a cocoon, the wasp pulled it out of the cell and threw 
it away. If it was a larva, she immediately brought in her Caterpillar but 
failed to lay an egg. Often she even captured some more caterpillars and 
stored them, still postponing the laying of the egg. It appeared that the pres- 
ence of a young larva stimulated the wasp to bring 1-3 caterpillars (corre- 
sponding with the second phase) and that an older larva stimulated the 
wasp to bring 3-7 caterpillars (corresponding with the third phase). 

Whereas, as we have seen, it is, the amount of food present at the solitary 
inspection visit which determines the wasp's behaviour in the second and 
third phase, the wasp is stimulated at her first visit by the age of the larva. 

Now the two series of experiments that served to investigate the part 
played by the solitary visit revealed the regulatory system, at work within 
the second or the third phase. They did not answer the questions as to the 
factors that bring the wasp from one phase into the next phase in the same 

Here the third series offers a suggestion. The age of the larva, which the 
wasp happens to find in her nest at her first visit, determines whether she 
will be brought into the second or into the third phase. This and other 
arguments, which cannot be treated in detail here, render it probable that 
the age of the larva has the same influence during the solitary visits. 

The following may be illustrated by the narrative of the activities of wasp 
GO. ... 

The wasp was marked shortly before she finished the third phase in nest 
A. After having closed this nest she paid a solitary visit to nest B. As the egg 
had not yet hatched, she closed this nest again without bringing a Caterpillar 
and began to dig a new nest C. Next morning she brought in a Caterpillar 
and laid an egg in this nest. Then she paid a solitary visit to B again where 
the larva had just hatched. Provisioning, therefore, is started (third phase) 
and is continued for 2 days. After she has finally closed nest B, the wasp 
pays a solitary visit to C. The egg in C has not yet hatched whereupon the 
wasp leaves C alone and*starts digging (D). As this was begun at a late hour, 
nest D was not completed before the end of the day. Next morning the wasp 
brought [sic] a solitary visit in C, nest D apparently being abandoned. In C, 
the larva has hatched and the wasp brought one Caterpillar. Then a new nest 
E was started, an egg was laid, whereupon, next day, the third phase in C is 
completed, which took 2 days. Next morning a solitary visit was paid to nest 
E where the larva had hatched. As a consequence one Caterpillar was 
brought soon afterwards. After that the wasp again began a new nest which 
is not taken into account here, because of incompleteness of my observa- 
tions. Next day the third phase in nest E was completed. 

It is on these observations that Baerends's diagram (Figure 20) is based. 
As the diagram illustrates, the hierarchical organization endows a 
sequence of behavior patterns with great plasticity and adaptability to 


IV. Complex Systems of Behavior Mechanisms 

changing environmental conditions—despite the fact that all its subsys- 
tems, as well as the interactions among them, are phylogenetically pro- 
grammed and modifications by learning play hardly any part in it at all. 
Modification by learning is restricted to the acquisition of local orienta¬ 
ron and otherwise has no influence on the stupendous adaptability of 
the hierarchical system. This makes it possible for the wasp to take care 
of four nests simultaneously, when each nest contains progeny at a dif- 
ferent stage of development requiring different actions pertinent to 
parental care. 

4. The Relative Hierarchy of Moods 

As already mentioned, a fixed motor pattern retains the character of a 
consummatory act, something striven for for its own sake by its own 
appetitive behavior, even if, in the sequence of a hierarchical organiza¬ 
ron, it plays the role itself of an appetitive behavior striving for the stim- 
ulus situation that will release the next link in the behavioral chain. 
Action-specific appetence for a behavior pattern can be so strong that its 
motivational power exceeds by far the stimulation coming from the next 
higher center, as represented in Tinbergen's diagram (Figure 19). One 
example is sufficient: A motor pattern common among ducks and geese, 
while they are swimming in shallow water, is that of "upending," that 
is, of stretching the neck straight down into the water and groping for 
food at the bottom. Usually this pattern forms a part of the system of 
feeding behavior and is motivated "from above" by an appetite for food. 
If, however, geese are kept on a pond that is devoid of vegetation, so that 
upending consistently fails to provide anything edible, the geese will 
still, nevertheless, from time to time, indulge in the motor pattern of 
upending for its own sake. Geese kept constantly in such an environ¬ 
mental situation and fed exclusively on corn that is strewn on dry 
ground and fed there to the extent that they refuse to eat any more, will 
begin again to feed when offered the same food thrown into the pond at 
the right depth for upending. In this case it is quite correct to say that 
the geese are now feeding for the sake of upending instead of upending 
for the sake of feeding—as is usually done. 

Another amusing example of two appetites swapping roles in this way 
was discovered by Ursula von Saint Paul and myself during our inves¬ 
tigaron of the ontogeny of "impaling" behavior in shrikes of the genus 
Lanius (1968). These birds hoard their prey, insects and small vertebrates, 
by impaling them on thorns. On an earlier occasion I had observed that 
in red-backed shrikes (Lanius collurio) the motor pattern of pressing the 
prey onto a thorn is innate, but that the orientation to the thorn must be 
learned. It soon became apparent that this lack of orientation was due to 

4. The Relative Hierarchy of Moods 


inadequate rearing, as mentioned in One/II/9. Saint Paul reared shrikes 
using a perfect technique; these birds oriented to a thorn at the first sight 
of it without any fumbling. Birds which had never been in contact with 
either a thorn or with prey for impaling were first offered a thorn in the 
guise of a nail of adequate size driven through one of their perches. The 
birds investigated the nail at once, examining it closely and then nib- 
bling it. Then they went in search of pieces of food to impale. Not find- 
ing any bits of food, they resorted to tiny and most inadequate objects. 
One greater shrike (Lanius excubitor) grabbed a small piece of dry butter- 
fly wing and carried it straight to the nail-thorn. The orientation of the 
impaling movement to the point of the nail was so exact that the bird 
succeeded in splitting this small object right through its middle—which 
may, of course, have been mere chance. In the complementary experi- 
ment, each of the inexperienced shrikes was presented with an object 
extremely suitable for impalement—a whole, dead, newly-hatched 
chicken. Each bird at once grabbed it and very obviously went in search 
of a thorn, wiping the object along the perches and on the bars of the 
cage and performing impaling motions in vacuo. Thus the presence of 
one of two objects belonging to the temporal sequence of the shrike's 
impaling behavior awakened appetitive behavior directed at the other, 
in spite of the fact that, in nature, grabbing prey must precede impale¬ 
ment on a thorn. 

Thus the temporal sequence in which activities follow each other in 
the normal performance of a hierarchical system need not necessarily be 
identical with the direction in which a "higher" center (in Tinbergen's 
diagram) is supplying excitation to a lower one. In the impaling behavior 
of our shrikes, as well as in the upending of our geese, the "eating drive," 
which might be assumed to be exerted by the highest center of the hier¬ 
archy, plays but an inferior role as a source of excitation, while motor 
patterns situated at the low level of consummatory acts exert a decisive 
influence on the animal's behavior. 

Among higher mammals, in whose behavior learning plays a much 
more important part than in that of birds, a much more complicated 
interaction takes place between the levels of a hierarchical organization. 
Paul Leyhausen, who investigated these phenomena in carnivores of the 
cat family (Felidae), speaks of a "relative hierarchy of moods" (relative 
Stimmungshierarchie). The acts of lying in ambush, stalking, catching, kill- 
ing, and finally eating prey, form a sequence which is obligatory only 
with regard to their common teleonomic function. Physiologically, each 
of the motor patterns involved retains the character of a consummatory 
act that possesses its own appetitive behavior independently of whether 
it is performed under the pressure of the higher level of tissue need or 
acted out in play, for its own sake. 

As has been explained in Two/II/4, Leyhausen has demonstrated that 


IV. Complex Systems of Behavior Mechanisms 

the endogenous ASP production of any motor pattern is strictly corre- 
lated in its quantity to the frequency with which the performance of the 
movement is needed in the daily life of the species. 

It is to be assumed that, even in felines, a primarily linear sequence of 
appetite and IRMs exists in ontogeny and that this sequence serves to 
guide the inexperienced kitten along the right route from stalking a prey 
to eating it. Leyhausen says: 

After the cat has caught, killed, and eaten several prey animáis and has thus 
experienced the connection between these three activities, it learns gradu- 
ally to substitute learned, or at least partially learned appetitive hahavior for 
the initial links of the chain of prey-catching motor patterns. Species as well 
as individuáis differ considerably in regard to the methods used in these 
learned movements. As in the case of the orientation of the killing bite, the 
fixed motor patterns themselves are in no way modified by learning (as are 
IRMs by the increasing of their selectivity), ñor is there any transformation 
of instinct through experience, as assumed by Bierens de Haan (1940). The 
fixed motor patterns remain autonomous and unchanged, side by side with 
newly acquired learned motor skills. In neurophysiology the existence of 
these two kinds of "motoric templates" has been known for a long time. 

5. The Locus of "Superior Command" (Übergeordnete 

As has been described in Two/I/13, when the abdominal nerve cord of 
the earthworm (Lumbricus) is separated from the supra-esophageal gan- 
glion, it continúes to send out the incessant coordinated impulses for the 
worm's creeping locomotion; the decerebrated spinal cord of a wrasse 
(Labrus) continúes to send out impulses that move the fish's fins until the 
death of the preparation. All tissues that are endowed with endogenous 
automatic production of excitation and that are normally subordínate to 
a central inhibiting influence, perform, on its removal, a continuous 
activity lasting as long as the preparation lives. Excitation-producing tis¬ 
sues of animáis that lack a central nervous system, such as the neuro- 
epithelia of coelenterates, represent intermedíate forms between ner¬ 
vous, muscular, and epithelial tissues and, as Batham and Pantin (1950) 
have demonstrated, are constantly and unceasingly automatically active, 
quite independently of any external influence. 

Whether excitation-producing tissues which are similarly independent 
exist in any organisms possessing a centralized nervous system is doubt- 
ful. It seems justified to suppose that, in higher animáis, excitation-pro¬ 
ducing organs are always subject to higher loci within the central ner¬ 
vous system which exercise those inhibiting, as well as those excitatory 
functions symbolized in Erich von Holst's analogy of the bridle and the 
spurs (Two/I/16). Even the sinus ganglion of the vertébrate heart is con- 

5. The Locus of "Superior Command" (Übergeordnete Kommandostelle) 


trolled by the nervus accelerans and by the nervus depressor cordis 
which perform these two antagonistic functions. 

The coelenterates investigated by Batham and Pantin are far from 
being the only organisms whose uncentralized nervous systems are 
devoid of superior, "commanding" loci. In sea urchins (Echinus), con- 
certed action of the peripheral organs, such as the ambulacral feet and 
the pedicels, is brought about by means of a direct mutual influence on 
each other and not via nervous pathways leading to a "center" and back 
again to the periphery. When a sea urchin moves hurriedly away from 
a starfish (Asterias), this movement is caused, anthropomorphically 
speaking, by a panic spreading among the spines. Jakob von Uexküll 
used to say: "When a dog runs, the dog moves its legs; when a sea urchin 
runs, the legs move the sea urchin." This assertion was based on the fol- 
lowing experiment reported by von Uexküll. A sea urchin was broken in 
half and the inner sides of both halves of the shell were scraped using 
sandpaper. The whole of the ambulacral system as well as the nervous 
system was thus completely removed. Then the two halves were joined 
together again by means of a spring clasp. The spines of the sea urchin 
still worked in coordinaron with one another. In this special case, the 
riderless horse of von Holst's parable does indeed exist; the sea urchin's 
reaction of fleeing from a star fish still functioned. And in this sense, von 
Uexküll's description of a sea urchin being a "reflex republic" is justified, 
provided one keeps in mind that the "reflex" no longer plays the all- 
important role ascribed to it during von Uexküll's time. 

The different roles performed by the loci of superior command are 
well exemplified by the functions of the central nervous systems in var- 
ious echinoderms. In sea urchins, the function of a superior commanding 
locus is hardly discernible by direct observation. In star fishes of the class 
Asteroidea, it is clear that messages are passed from one arm to the other 
arms. For instance, the animal finds its way back into sea water imme- 
diately, walking on its ambulacral feet, if one of its arms touches the 
water surface with its tip. What is so remarkable in this is that the star 
fish "walks" by means of coordinated movements that, necessarily, must 
pulí at different angles to the longitudinal axis of each arm. 

If the star fish is turned over onto its back, a coordinated movement of 
its arms serves to right it. In brittle star fish, Ophiroidea, the central ner¬ 
vous system can effect coordinated, undulating movements in the arms 
themselves, so that the animal "walks" much more quickly than any 
other star fish can and in any "desired" direction, which generally means 
toward cover. 

Some Holothuraidea have developed into virtually bilateral animáis, 
particularly the genus Synapta which lacks ambulacral feet and "walks" 
by means of peristaltic movements of its entire body. It certainly is the 
most lively of all echinoderms and, in its behavior, gives the impression 
of a bilaterally symmetrical animal. 


IV. Complex Systems of Behavior Mechanisms 

All true bilaterally symmetrical animáis, in other words, all those 
animáis having a front end and a rear end and, therewith, a preferred 
direction of locomotion, possess, near their front ends, a central com- 
manding locus as part of their nervous system, that is, a "brain." This is 
also the case even in those animáis not possessing a "head," as, for 
instance, clams (Lamellibranchiata). The "invention" of a "head," that is, 
of a concentration of sense organs and nervous tissues at the end, which 
precedes during locomotion, brought with it such obvious teleonomic 
advantages that the invention has been made twice, independently, dur¬ 
ing the history of animáis. Arthropods, (crustácea, centipedes, insects, 
arachnomorphs) as well as mollusks, may have inherited their heads 
from their mutual ancestors, the annelid worms. In all members of this 
great taxon, the mouth develops out of the blastopore, which means that 
it is homologous to the "protostome," the single body apperture of 
coelenterates, wherefore these creatures are termed 'protostomata.' Ver- 
tebrates and a few others have a mouth of different origin, and are 
termed 'deuterostomata.' Exaggerating for effect, it can be said that ver- 
tebrates and insects have their heads at opposite ends of their bodies. 
Both protostomes and deuterostomes possess a surprisingly similar, if 
only analogous organization of the head; nobody could ever have a 
moment's doubt about where a wasp or a squid has its head, its eyes, or 
its mouth. The functions of the two kinds of independently evolved 
brains also appear analogous. 

A superior command locus obviously becomes necessary at the stage 
of evolution when an organism has developed more than one system of 
behavior patterns of which only one can function at any given time. 
Such a juncture becomes imperative in order to prevent incompatible 
motor patterns from "meshing gears" by being discharged simultane- 
ously. An earthworm (Lumbricus) can creep forward and backward; it can 
eat, roll up dead leaves, and draw them into its burrow; it can copúlate, 
draw back with lightening speed into its burrow, and, finally, it can— 
this I have never seen—move away quickly, on the surface of the 
ground, by executing a vertical undulating movement. As simple obser¬ 
varon confirms, each of these movements is performed by itself, neatly 
separated from and never combined with any of the others. In an epilep- 
tic fit, mixed discharges can occur pathologically. This demonstrates that 
such a dysteleological occurrence is possible, in principie, and must be 
prevented by special mechanisms. A comparatively simple physiological 
organization achieving this purpose is what biocyberneticists cali a máx¬ 
imum selecting system. It is represented in Figure 21. The mechanism is 
built in, like a filter, between the several sources of specific excitation 
and the motor pathways through which ASP is discharged. By causing 
a "backward-directed subtraction," it effectuates a simple mutual inhibi- 
tion among the action patterns involved. The principie of "lateral back¬ 
ward inhibition" makes certain that the behavior pattern possessing, at 

5. The Locus of "Superior Command" (Übergeordnete Kommandostelle) 



Figure 21. Functional diagram representing the reciprocal inhibition among 
behavior tendencies in accordance with the principie—plausible for such cases— 
of retroactive inhibition. Respiration is usually not included in this system; 
breathing continúes unaffected by other forms of behavior. (Czihak, Langer, Zie- 
gler in Biologie.) 

the moment, the highest ASP valué has free access to its consummatory 
action unobstructed by any competing motivation. 

It can be hypothesized that the primary function of the simplest pos- 
sible brain-like organs of bilateralia has been based on two functions. 
The first is that of a máximum selecting system serving to pass motor 
patterns one at a time; the second is that of IRMs serving to decide which 
of these behavior patterns should be discharged in a given situation. 

All motor patterns not in use at a particular moment are kept under 
constant restraint by central inhibitions. It has been demonstrated by 
Erich von Holst, Kenneth Roeder, and by many others, that in annelid 
worms, in insects and in vertebrates, superior command loci within the 
nervous system are exerting a constant inhibitive influence that prevenís 
the subordínate processes of endogenous impulse production from hav- 
ing a continuous influence on motor activity. The same functional prin¬ 
cipies prevail in the human brain, too, as is demonstrated by the multi- 


IV. Complex Systems of Behavior Mechanisms 

farious effects of brain damage. In Parkinson's disease, as well as in the 
so-called post-encephalitic syndrome, the action of disinhibited motor 
processes form the predominant symptoms. 

Kenneth Roeder, who with his co-workers investigated the function 
of central inhibitions in insects (1960) says: 

Regions of the nervous system responsible for the inhibitory effect were 
located in the following manner. Removal of the compound eyes, ocelli, and 
antennae of the cockroach caused no permanent increase in the efferent out- 
put of the last abdominal ganglion [that is, the centrally produced and coor- 
dinated impulses for the specific motor patterns of the phallomeres of the 
copulatory organ in the last segments of the insect's body]. Removal of the 
brain (supra-oesophageal ganglion) was followed by an increase in small 
spike activity in 4 out of 8 preparations. Decapitation (or transection of the 
cercival [sic] connectives) was followed by a much more pronounced 
increase that included volleying in the large libres. The effect was similar to 
that caused by transection of the abdominal nerve cord. This shows that, as 
in the mantis, the sub-oesophageal ganglion is the main source of the inhib¬ 
itory effect. In some cases transection of the abdominal cord subsequent to 
decapitation brought about a small additional increase in the efferent activ¬ 
ity, suggesting that the thoracic ganglia may also contribute in smaller mea- 
sure to this inhibitory effect. 

Assuming that suppression of this efferent activity [copulatory motor pat¬ 
terns] in the intact nervous system is due to a tonic discharge of inhibitory 
impulses in descending nerve fibres, it is somewhat surprising to encounter 
a delay of as much as 15 min between removal of the inhibitory effect by 
cord transection and the onset of efferent activity. Efforts were made to find 
out whether the duration of this delay is determined by the site at which 
the inhibitory pathways are interrupted. There was no evidence that the 
delay differed when transection of the cord was made in the neck as com¬ 
pared with the abdominal región. However, since the delay was so variable, 
ranging from a few seconds to as much as 15 min, its duration in relation to 
the site of transection could have been obscured in the small number of 
observations that have been made. 

The method of cord transection permitted only one release of endogenous 
activity per preparation, so attempts were made to establish a reversible 
block of the inhibitory effect. An electronic method was first tried. The pas- 
sage of direct current through the nerve cord via an anode placed on the 
cord pitted against an indifferent cathode brought about a small increase in 
the efferent activity in nerves IX and X in 3 out of 11 cases, but this action 
was not sustained for the duration of the polarizing current. In the other 
cases there was no change in efferent activity. In 6 of the 11 cases the nerve 
cord was transected subsequent to treatment with the polarizing current and 
in each of these there was the usual increase in efferent activity. 

Localized ionic block of descending inhibitory pathways in the nerve cord 
was attempted through the application of isotonic potassium chloride and 
sodium-free (sucrose substituted) Hoyle's saline. Isotonic potassium chloride 
applied locally to the intact nerve cord takes about 30 min to block conduc- 

5. The Locus of "Superior Command" (Übergeordnete Kommandostelle) 


tion in the ascending giant libres, but if the sheath surrounding the cord is 
removed before potassium chloride application the block occurs in a minute 
or so (Twarog and Roeder, 1956). These agents were applied to intact and 
desheathed portions of the abdominal nerve cord while monitoring the out- 
put of efferent activity in nerves from the last abdominal ganglion. Diffi- 
culties were encountered in restricting the potassium chloride to a limited 
región of the nerve cord, and attempts were made to prevent it from reach- 
ing the last abdominal ganglion by submerging the latter in mineral oil. Of 
7 experiments in which efferent activity continued in the last abdominal 
ganglion throughout the rather involved procedure 1 showed no change, 4 
showed a slight increase not comparable with that achieved by cord tran- 
section, and in 2 there was an immediate increase followed by cessation of 
all efferent activity. In the latter activity was restored after 5 min by washing 
with saline, and the cessation could have been due to leakage of potassium 
chloride to the región of the ganglion. 

The failure of these two attempts to block the inhibitory action caused 
some surprise, and is interesting in itself. Three possible explanations come 
to mind. First, the methods were technically inadequate. However, the same 
method of producing a potassium chloride block was successful with the 
giant fibres (Twarog and Roeder, 1956). Second, the inhibitory fibres have a 
small diameter and are in some way protected from agents that block the 
giant fibres. Third, the inhibitory action of the higher centres is exerted by 
agents other than nerve impulses. The last two possibilities remain entirely 
open, and the mode of inhibition will be considered further in the 

In higher vertebrates, the "superior command locus" has a much more 
complicated function than those investigated by Roeder and von Holst, 
as it has to act as a connecting link between phylogenetically pro- 
grammed and individually acquired behavior. Also, its function is quite 
certainly not limited to merely inhibiting unwanted activities; it obvi- 
ously acts just as frequently as the initiating source of activity in general. 
It uses the spur as often as the bridle. Although "multi-purpose" activi¬ 
ties do possess a suíficient and even an abundant supply of endogenous 
impulse production (Two/I/12) there is no doubt that quite often a very 
energetic active impulse emanates from superior command loci—I have 
been trying to avoid the word "center." As is to be explained later in the 
chapter on motor learning (Three/IV/1-3), the so-called voluntary 
movements have especially evolved in order to be at the beck and cali of 
active superior command loci, and to be built into complicated systems 
in which not only learning, but also "insight" in the form of exploiting 
instant information takes a decisive part (Two/VI/12). 

Even though in the highest animáis and in man the superior command 
loci still retain their primary function of keeping motor patterns under 
inhibition and of liberating them only at the right moment, direct obser¬ 
varon makes it obvious that, additionally, the very highest loci of most 


IV. Complex Systems of Behavior Mechanisms 

warm-blooded animal's brains are spontaneously active as long as the 
animal is awake. Unlike a worm or a snail, a lively creature such as a 
songbird or even a small coral fish is constantly moving receptor organs 
in different directions, actively searching for stimuli during the hours of 
its wakeful activity. It is primarily in organisms of this type that the State 
of wakefulness is clearly distinct from that of sleeping. 

This wakeful searching for stimulation can be regarded as a precursor 
of true exploratory behavior, to be discussed in Three/VI. A stickleback 
while remaining motionless in the water moves its eyes about, "feeling 
out" its surroundings prior to moving in an appropriate direction. The 
same is true for many small songbirds, whose eyes are never still for a 
moment—as every photographer comes to realize with regret. In man, 
the movements of the eyes, focusing on one point of the ambient envi- 
ronment after the other, are so continuous that their cessation immedi- 
ately draws our attention: we then say that a person "stares into space," 
which expresses the fact that these movements serve to maintain spatial 

This activity certainly is based on endogenous impulse production 
rather than on reflex processes. It is also an obvious assumption that 
superior commanding loci of higher vertebrates perform an exciting as 
well as a bridling effect on the subordínate processes. A strong argument 
in favor of this assumption is that the voluntary movements which are 
known to exert the strongest influence in driving multi-purpose activi- 
ties are, at the same time, more directly obedient to superior commands 
than any other kind of neural process, in fact, obedient until complete 
exhaustion of the organism as a whole has set in. 

Chapter V 

How Unitary Is "An Instinct"? 

1. The Danger of Naming Instincts by Their Functions 

A "system," according to Paul Weiss's somewhat aphoristic definition, 
"is everything that is unitary enough to deserve a ñame." In choosing 
any ñame one must avoid following the example of medieval pseudo- 
science which confused effect and cause. When air was streaming into a 
vacuum, a "horror vacui" was made responsible for that effect; the "phlo- 
giston" was assumed to explain burning, and so forth. In exactly the same 
way, vitalistic psychology, rampant at the turn of the century, used terms 
such as the "escape instinct," the "reproductive instinct," and even the 
"instinct of self-preservation," deeming this to be a sufficient explanation 
of behavior. This seemed legitimate as long as an "instinct" was regarded 
as a preternatural factor, neither standing in rieed of ñor accessible to a 
causal explanation. 

To cali a function by a ñame stemming from its teleonomic effect is, in 
itself, permissible, provided that one remains aware of the danger that 
this ñame may, as John Dewey pointed out, insidiously raise the false 
pretensión of being an explanation for the function it describes. When 
dealing with a complex behavior system fulfilling a unitary teleonomic 
function, we are justified in naming it according to its function and to 
speak of a "reproductive instinct," or an "aggressive instinct," as both 
Tingergen and I have done (1938), but when doing this we must never 
forget that the system thus embraced by a functional concept represents 
a very loosely tied unit within which the teleonomic cooperation of very 
many autonomous parts is calling, if not screaming, for a causal expla¬ 
nation. The ñame, far from furnishing such an explanation, is nothing 
but a challenge to physiological analysis. The justification for providing 
a ñame depends exclusively on the success of this analysis. 


V. How Unitary Is "An Instinct"? 

Even if one remains fully aware of all these facts, it is still dangerous 
to ñame a somewhat larger behavior system by its function. Even though 
an expression like "reproductive instinct" or like "aggressive instinct" 
no longer implies postulating a preternatural factor, remnants of teleo- 
logical thinking still tend to suggest, at least to some, the assumption of 
a single, if natural, source of causation. The reproach of assuming a 
"mono-causality" of instinct, particularly of aggressive instinct, has often 
been raised against me, although in my book on aggression I devoted an 
entire chapter, "The Great Parliament of Instincts," to the way in which 
a multiplicity of independent motivations are interacting in aggressive 

2. The Multiplicity of Motivations 

Tinbergen's diagram of the hierarchical organization of instincts (see 
Figure 19b) tends to further the conception of a one-way causation, but 
quite erroneously, since no one was or is more acutely conscious than 
Tinbergen that innumerable important feedback causations stem from 
the lowest levels of the hierarchy and take effect at the highest ones. As 
Heiligenberg and others have shown, a high pressure exerted by the ASP 
of one motor pattern, even at the lowest rung of a hierarchy, demonstra- 
bly exerts a decisive influence on processes taking place at the highest 

In his 1959 article, "Unitary Drives," Robert Hinde has convincingly 
argued against the assumption of a mono-causality of instincts and 
against the concept of unitary drives. "One feature of behaviour which 
drives are postulated to explain is the temporal correlation between 
related activity. Most behaviour consists of sequences of activity and it is 
often suggested that these occur together, because they are governed by 
the same drive." Neither Tinbergen ñor Baerends ever thought so; both 
saw very clearly that the teleonomically correct temporal sequence of 
activites within a hierarchical organization is programmed by its struc- 
ture. In other words, the sequence is determined by the fact that each 
temporally preceeding link of appetitive behavior is switched to the suc- 
ceeding one by one particular IRM. The function of each of these IRMs 
has been experimentally demonstrated by both authors. 

Robert Hinde has investigated the multiplicity of the motor patterns 
which are integrated into the unitary function of the nest building of 
canaries (Serinus canaria). He showed that the single motor patterns are 
very independent of one another and, on the basis of this fact, Hinde 
argued correctly against the assumption of their being caused by a uni¬ 
tary drive. However, he ignores the fact that every single one of the 
motor patterns concerned in nest building possesses its own appetitive 
behavior as well as its own IRM. He also refuses to accept the fact that 

2. The Multiplicity of Motivations 


every motor pattern is a systemic unit which in itself is much more 
closely integrated and forms a much better defined unit than the super- 
imposed, much more loosely connected hierarchical system which, fol- 
lowing Tinbergen, we are wont to cali "an instinct." Hinde avoids our 
assumption of an action-specific potential (ASP) which can be accumu- 
lated and exhausted. For this, an unavoidable assumption in my opinión, 
he substitutes the following argumentation: External stimuli can exert a 
specific as well as an unspecific influence on behavior. Unspecific stim- 
ulation is necessary to sustain a sufficient measure of general nervous 
excitation, while specific stimuli determine the intensity of behavior. 
According to Hinde, it is sufficient to assume a continuous general activ- 
ity within the nervous system, while there is no necessity to postúlate 
specific motivations of behavior. 

I fully agree that unspecific stimulation is necessary to keep up an 
adequate measure of general excitation (Two/I/8). However, I strongly 
object to the second assertion that the intensity of response is dependent 
only on specific stimulus situations. Hinde refuses to acknowledge not 
only the banal, ordinary, everyday experience of anyone who knows 
animáis, but also the irrefutable results of the experiments made by A. 
Seitz (1940, 1941), H. W. Lissmann, D. Franck and U. Wilhelmi (1973), 
and many others. I refer to what has been said in Two/I/5 on the method 
of dual quantification. 

Hinde's failure to acknowledge these facts is all the more surprising 
since he correctly, and in detail, describes the several fixed motor pat- 
terns of nest building and does not neglect the stimulus situations which 
release them. In doing this he cannot help mentioning spontaneity, 
appetitive behavior, and the threshold fluctuations concerned with these 
elements of behavior. Nonetheless, he concludes, "... are we then to 
postúlate a sepárate drive for each of these activities? If so, where do 
these processes stop, for each of these activities can be analyzed into con- 
stituent movements?" 

The first of these questions has been answered correctly by Leyhausen: 

The extent of the complexity of the constituent movements in question is 
quite inessential. If a pattern can appear for itself, independently of its con- 
text in the whole of the teleonomic sequence and if, furthermore, it is striven 
for by a specific appetitive behavior directed at its releasing situation, these 
very phenomena are its own independent motivation, it is not necessary to 
postúlate one! 

The second question concerning the level of integration, at which our 
search for a unit of motivation has to stop, receives an equally clear 
answer. It has to stop at the level of those patterns which show the phe¬ 
nomena of threshold-lowering, intensity fluctuations, in short, all the 
phenomena of spontaneity discussed in Chapter Two/I. All these phe- 


V. How Unitary Is "An Instinct"? 

nomena are not shown by single muscle contractions or even by move- 
ments of whole limbs, in other words, not below the third level, that of 
the consummatory act in Tinbergen's diagram. The fact that it is the unit 
of the fixed motor pattern which has its own motivation is just what 
makes its discovery by Whitman and Heinroth so very important. The 
whole Science of ethology is based on this discovery. 

The multiplicity of motivations entering into the functional entity of 
a so-called instinct corresponds to the number of independent motor pat- 
terns taking part in the system. The analysis of these many independent 
motivations is indeed quite as important as Hinde has urged. Our aim, 
however, is not the elimination of the concept of a unitary and teleon- 
omically functioning system of autonomous motivations, but the phys- 
iological analysis of their interaction which makes the system a unit. The 
idea of a hierarchical organization of instincts as developed by Tinber- 
gen, Baerends, and Leyhausen may represent a great simplification of 
reality, but it uneqivocally shows the way which causal analysis has to 

The methods of approach dictated by these considerations are impor¬ 
tant even in those cases in which the functional, teleonomic unity of the 
single motor pattern is unquestionable, though a hierarchical temporal 
sequence cannot be shown. One behavior pattern of nest building, inci- 
dentally not described by Hinde in the domesticated canary and more 
familiar to me in some of the bird's wild relations (Carduelidae), consists 
in salivating on nesting material, treading on it with one foot and trying 
to fasten it onto branches by a certain complicated movement of bilí and 
tongue. Simultanously with this motor pattern, female Carduelidae show 
a second pattern, the function of which is the molding and smoothing of 
the nest cup. It consists of simultaneously pushing forward the chest by 
kicking backward with both feet and pressing the shoulders of the wing 
outward. This pattern is obviously very oíd; there does not seem to be a 
carinate bird in which it is lacking. Among the small songbirds here 
under discussion, this motor pattern has acquired another function. By 
performing the movement in a crotch of branches chosen as a potential 
nesting site, the bird acquires information about how favorable or unfa- 
vorable the locality in question is. The greater number of branches which 
the bird touches in performing the cup-molding pattern, the more favor¬ 
able the location obviously is for the building of a nest cup, the shape of 
which is largely preformed and supported by the branches touched. 
Learning processes, presumably in the form of conditioned appetitive 
behavior, play a role in the choice of a nest site, since the females of 
many tree-nesting birds regularly try out different possibilities by per¬ 
forming the cup-molding pattern before they begin to build. On the 
other hand, and in contrast to the nest building of rats (One/II/9), learn¬ 
ing does not seem to improve the temporal sequence of the motor pat- 
terns used in building a nest. 

3. Integrating Effect of the Instinct Hierarchy 


3. Integrating Effect of the Instinct Hierarchy 

Tinbergen definitely underestimates the valué of his diagram (see Figure 
19b) when he says that his hypothesis of a hierarchical organization of 
instincts represents nothing more than a diagram designed to bring some 
provisional order into our theorizing. This statement might make us 
underrate the number of experimentally proved facts explained by the 
hypothesis. These facts are: 

1. The number of IRMs taking part in the system. 

2. The specific stimulus situations releasing each of these IRMs. 

3. The form and function of the motor patterns released. 

4. The interrelations of these behavior patterns, the mutual facilitation 
and inhibition as indicated by the arrows and double arrows. 

5. The teleonomic function of these experimentally analyzed processes 
which consist in creating adaptive sequences out of motor patterns 
rigid in themselves. 

In spite of the unavoidable simplification which comprises primarily 
an omission of the manifold feedback effects contributing to the func- 
tional integration of the system, Tinbergen's diagram nonetheless shows 
very clearly in what way the interaction between its parts serves the 
integration of the whole. It makes sense from the point of view of teleon- 
omy that, for instance, a male stickleback is not able to develop nest- 
building motivation before having satisfied appetitive behavior of the 
first order by reaching shallow, warm water with vegetation. Even more 
striking is the adaptive order-producing function of the sequence of dif- 
ferent actions in the system of nest building and parental care which 
Baerends investigated in the diggerwasp (Ammophila) (1941). 

As has already been said, one fixed motor pattern can constitute the 
appetitive behavior which aims at the release of a subsequent one. 
Henee, one motor pattern can receive motivational energy from another 
one. Conversely, driving power can also be exerted from the higher level 
in the direction of the lower in the sense of the arrow in Tinbergen's 
diagram (see Figure 19b). As discussed in Two/IV/4 in connection with 
the relative hierarchy of moods, the relationship between action patterns 
that are driving and those that are being driven can reverse itself accord- 
ing to the ASP of the several action patterns, as well as to the general 
stimulus situation and the State in which the organism finds itself at the 
moment. This reversibility helps to achieve an even greater adaptive var- 
iability of the system. 

Furthermore, learning processes in higher animáis can act as another 
integrating factor by setting a common goal to a great number of behav¬ 
ior patterns taking part in a hierarchical system. Highly adaptive "short 
cuts" to the teleological end can thus be achieved. 


V. How Unitary Is "An Instinct"? 

4. Interaction Between Motor Patterns 

As indicated by the horizontal double arrows in Tinbergen's drawing 
(see Figure 19b), motor patterns on the same hierarchical level tend to 
inhibit each other. Obviously a stickleback occupied at the moment with 
fighting cannot simultaneously build a nest or court a female. Motor pat¬ 
terns of the same hierarchical level can be based on a common State of 
readiness characteristic for their level, but they are prevented from 
breaking out simultaneously by a specific mechanism which is akin to 
that of the máximum selecting System mentioned in Two/IV/5. Simple 
superposition of the endogenous impulses stemming from two motor 
patterns, as will be described in Two/VII/2, would obviously be unadap- 
tive in this case. 

If one wants to decide, by statistical evaluation of observational data, 
whether mutual inhibition or mutual facilitation prevails between two 
motor patterns, the answer would depend on the time span one chooses. 
If one judges from samples taken only seconds distant from each other, 
the result will be a relationship of absolute mutual inhibition between 
motor patterns lying on the same level. If, however, one chooses samples 
taken one or several days apart, it will appear that these motor patterns 
are activated simultaneously and by the very same motivations. This cor- 
responds generally with what has been said in Two/1/4 on the greater 
inertia of change characteristic for the "moods" of higher and lower 

The closeness of two motivations within a hierarchical System finds its 
expression in the time that must elapse before the organism is able to 
change from one motor pattern to the other. In the stickleback, rival 
fighting and courtship can follow each other within a few seconds. In 
cichlids, in Hemichromis bimaculatus for example, the high general exci- 
tation elicited by fighting can be channeled directly into a correspond- 
ingly high excitation in courtship. After having been mated for a long 
time, oíd specimens of this species show courtship activities only of a 
remarkably low intensity. Even spawning and fertilization are per- 
formed as a matter of routine; the lack of excitation gives the impression 
that the entire performance is running along a path smoothed by habit. 
If one confronts such a tired and long-married couple with a highly col- 
ored fellow member of the species, both mates will attack it furiously, 
and it is quite surprising that in the general mix-up the mates hardly 
ever hit each other rather than the intruder. If the latter is now suddenly 
removed, both mates, still in high fighting mood, are for a moment in 
obvious danger of attacking each other, but immediately afterward their 
excitation is channeled into the path of sexual behavior and the fish per- 
form courtship movements of the highest intensity such as they have not 
evinced since the first phases of pair formation at the time of "young 
love." By a more exact analysis of the motivations underlying agonistic 

5. Motor Patterns Not Specific to the System 


and sexual motor patterns we are confirmed in our supposition that both 
have similar conditions of internal hormonal readiness. 

Generalizing, it is permissible to say the following: The time necessary 
to change from the readiness for a certain action pattern to that of 
another becomes the longer the higher the levels of the readinesses in 
question are placed within the hierarchical organization embracing both. 
Should one try to switch the behavior pattern of a pair of Hemichromis by 
changing their external stimulus situation—not, as described above, 
from fighting to courtship, but from an escape reaction of corresponding 
intensity to courtship or to agonistic behavior—one would find that the 
fish requires not seconds, but nearly half an hour to perform the neces¬ 
sary change of "mood." 

In the experiments of electrical brainstem stimulations performed by 
Erich von Holst on chickens, he encountered the same phenomenon. A 
motor pattern of low integrational level released by itself can very 
quickly disappear or be replaced by another. If, for instance, of the whole 
system pertaining to alarm and flight, von Holst released only one motor 
pattern belonging to the lowest level of integration, for example the rais- 
ing of the head and looking about in all directions, this motor pattern 
disappeared almost simultaneously with the ending of the stimulus and, 
in an equally short space of time, the chicken was ready for motor pat¬ 
terns belonging to other systems. If, however, the electrode was posi- 
tioned in a way to activate the whole alarm-and-escape system, the 
arousal of which actually begins with the motor pattern just mentioned, 
but, with stronger stimulation, escalates to uttering the warning cackle 
and, finally, to flying away, the bird, after the cessation of the stimulus, 
takes much longer to quiet down, let alone to perform behavior patterns 
belonging to an altogether different system. The larger the system acti- 
vated and the higher the level of its activation, the longer it takes for the 
disappearance of the inhibiting effect which its excitation exerts on com¬ 
parable neighboring systems. 

The results obtained by Erich von Holst agree very well indeed with 
the hypothesis developed by Tinbergen and Baerends concerning the 
hierarchical organization of motor patterns belonging to "an instinct." It 
may be, however, that the abnormally strong stimulation of the system 
investigated causes an apparent "mono-causality" which is not normally 
characteristic of the function of the system in the intact animal. In any 
case, the findings of von Holst do not constitute an argument for the 
mono-causality of instinct. 

5. Motor Patterns Not Specific to the System 

There is a particular, rather well-defined type of fixed motor pattern 
which can be built into very different systems, into hierarchical ones as 


V. How Unitary Is "An Instinct"? 

well as into less integrated mosaics of behavior. As discussed in One/I/ 
12, describing the way in which all fixed motor patterns can exert a driv- 
ing function just as well as they can allow themselves to be driven by 
other motivations, I have already explained that there are certain multi- 
purpose activities, for example, fixed motor patterns that can be used for 
different purposes and activated by very different neural mechanisms. 
The same principie exists at lower levels of neural integration: the same 
retinal element can simultaneously act as a member of several different 
organizations. As Lettvin and his co-workers have shown, one cell in the 
retina's layer of ganglia can subsume the report of several receptor cells 
and abstract from it a definite report and convey it towards the center. 
One reports only a simultaneous change from light to darkness and the 
reverse (the so-called on-off effect), others abstract more specific mes- 
sages, for instance, the progression of a dark, convex contour in a direc- 
tion from the left to the right side, and so forth. On the highest level, a 
man can be a member of several organizations, of a club, an army, and 
of a political party. 

Some of the "multipurpose actions" here under discussion are found 
built into not only several, but practically into all the more highly inte¬ 
grated systems. Locomotion, for instance, forms an indispensable part of 
most existing hierarchical organizations of instinct. The existence of mul¬ 
tipurpose patterns preeludes our defining "an instinct" through an enu¬ 
meraron of the motor patterns functioning within its context. 

Even motor patterns that have very obviously originated in the Service 
of a quite particular system which, in turn, serves an equally specialized 
function, can be phylogenetically transferred, in an apparently rather 
sudden change of function, into an altogether different system perform- 
ing an altogether different teleonomic function. Teeth and jaws have 
evolved to catch and eat prey; the "invention" of using these organs for 
self-defense, though seemingly obvious, has not "occurred" to all gna- 
thostome vertebrates. Very few teleost fish try to bite when caught or 
cornered, even if they possess a great gape with sharp teeth; neither pike 
(Esocidae) ñor salmón (Salmonidae) do. Among the great family of blen- 
nies, only the giant Anarrhichas lupus, called the "sea wolf" in Germán, 
bites most vigorously, and so does the barracuda (Sphyraena barracuda) 
and the characin Hoplias malabaricus, which in its external appearance and 
ecology is very like a pike. The heavily armed piranha, also a member of 
the order of Characinidae, apparently does not bite in self-defense. All 
Plecostomidae, in particular the large triggerfish, bite most viciously. It 
seems not to be known whether sharks bite during rival fighting; some, 
but by no means all, bite when caught by the tail. Very few amphibians 
bite; only the horned toad (Ceratophrys) does. And even among reptiles 
there are many, such as the grass snake (Natrix natrix), which have not 
grasped that hunting weapons can be used in self-defense. Young alli- 
gators (Alligator mississippiensis) do not bite, while young crocodiles (Cro- 

6. Chapter Summary 


codilus niloticus) not only bite, but at the same moment start to twist side- 
ways—as I have learned the hard way. 

We know two kinds of fish, quite unrelated to each other, which have 
a very special organization of organs and motor patterns that serves pri- 
marily to scrape algae from the substratum, but has secondarily been 
transferred into the behavior system of rival fighting. Both the "kissing" 
gourami (Helostoma temmincki) and the blenniid Escenius have jaws and 
teeth adapted to scraping, and each of the two contending fish "sarapes" 
at the mouth of the other with exactly the same motor pattern. In the 
gourami, this gives the impression of passionate kissing. 

In the above, the expressions I employed to describe a motor pattern 
which "originated in the Service" of one system and later was "used by" 
another could give the impression that these "multipurpose" activities 
tend to be motivated exclusively by the influence of the higher levels, or 
"centers." This certainly is not the case. Even the activities of locomotion, 
more susceptible than any of the others to being driven by motivation 
coming from "above," contribute by their autonomous production of 
impulses to the excitability of the whole system. In the cichlid fish (Pel- 
matochromis subocellatus kribensis), Heiligenberg has shown that the readi- 
ness to move exerts a decisive influence on the system of escape behavior 
(1964). The same standardized stimulus causes the fish to dart away if it 
possesses, at the moment, a high readiness to swim, yet when impinging 
on an individual which, at the moment, possesses a low readiness to 
move, the stimulus will cause the fish to develop cryptic coloration and 
to hide. Anyone who has ever ridden a horse that has not been out of 
the stable for a long time is aware of the way in which the "damming 
up" of locomotor activities can influence the entire behavior of an 

6. Chapter Summary 

1. Under the ñame of "an instinct" or "a drive" we conceive a spontan- 
ously active system of behavior mechanisms sufficiently connected by 
a common function to deserve a ñame. Choosing a ñame for such a 
system corresponding to its function should not be misinterpreted. 
We neither believe in an extranatural factor guiding the organism to 
a teleologically determined goal, ñor do we believe that there is a sin¬ 
gle "mono-causal" physiological process which is responsible for the 
spontaneity of the system. 

2. Systems of instincts, as we understand them, are always activated by 
a multiplicity of motivations independent of each other. The best 
known of these motivations are the appetites for the performance of 
the several fixed motor patterns participating in the system. Any 
motor pattern that, independently of the teleonomic context of the 


V. How Unitary Is "An Instinct"? 

system, can be shown to possess its own spontaneity and to cause 
appetitive behavior directed at its release, is a motivation. 

3. A number of independently spontaneous motor patterns and innate 
releasing mechanisms (IRMs) are integrated into a functional whole 
by the structured program governing their interaction within the 
hierarchical organization of the "instinct." Tinbergen, Baerends, and 
Leyhausen have demonstrated the structure of this program by con- 
clusive experiments. 

4. These interactions consist partly of mutual inhibition, partly of 
mutual facilitation, and often in obligatory temporal sequence, one 
motor pattern bringing about the releasing situation for the next. 

5. The unit of an instinct cannot be defined by enumerating the motor 
patterns taking part in it, because many of them can be built into, or 
"used by," several independently functioning systems. Although 
these "multipurpose" activities can be driven by strong impulses 
stemming from higher strata of the hierarchy, they represent, by their 
autonomous production of excitation, an essential motivating power 
for the whole system. 

Chapter VI 

Mechanisms Exploiting Instant 

í. Receiving Information Does Not Always Mean Learning 

The fact that an organism receives information does not imply uncon- 
ditionally that it learns something, although of course the receiving of 
new information is an indispensable prerequisite for learning. As will be 
discussed in the third part of this book, learning, in the widest possible 
meaning of the word, is defined as an adaptative modification of behavior, 
in other words, an improvement of the physiological "machinery" 
whose function is behavior. As in any other adaptive process, adaptation 
to a certain given in the organism's environment invariably means that 
information about this given must somehow have been fed into the 
organic system. In phylogeny this is achieved by the age-old, trial and 
success method of random genetic change and subsequent selection, but 
in ontogeny this is achieved by learning. Both processes have one faculty 
in common; both can acquire and store information. The first process for 
gaining information is as oíd as life itself; the second could only come 
into being after a more or less centralized nervous system had evolved. 

In addition to and independent of these two mechanisms for acquiring 
information, there exists a third great category of processes, similar to 
the two just mentioned with regard to the faculty for gaining and for 
exploiting information, but quite unlike them in another important 
aspect: the processes of this category cannot retain the information; in 
fact, they must not, because they must maintain a readiness to counter- 
mand any of their own prior messages at a moment's notice. 

While we are walking we are receiving and exploiting, at any single 
moment, a vast amount of information. Our proprioceptors keep us 
informed about what our legs are doing, exteroceptors report the prop- 


VI. Mechanisms Exploiting Instant Information 

erties of the ground on which we are stepping, our gravity and acceler- 
ation receptors keep telling us the position of our center of gravity in 
relation to our support, by optical means we are informed about the 
speed of our movement, optokinetic responses keep us going straight 
and tell us about any deviations, and so on. Every single piece of infor¬ 
maron thus received is evaluated and exploited, to be erased immedi- 
ately afterwards. The messages must not leave any vestiges whatsoever— 
so that they cannot impede any contrary response which may be 
demanded at the very short notice of a hundredth of a second. 

Mechanisms exploiting instant information are omnipresent. An adap- 
tive change of behavior achieved by the method of genetic change and 
selection requires, as a minimum amount of time, the duration of one 
generation. An unimaginable environmental constancy would be 
required for the survival of organisms devoid of mechanisms enabling 
them to adjust to any small deviation from the norm of external condi- 
tions, and thus to keep constant the functions of their internal processes. 
Regulating mechanisms must have come into existence when life did. 

Many students of behavior are of the opinión that "learning" takes 
place during every type of complex behavior. This error is probably 
based on their confusing the function of learning with that of the mech¬ 
anisms exploiting instant information. Perhaps this confusión is fur- 
thered by the ubiquitous expression, "I have just learned that. . .," mean- 
ing "I have just been informed." There are innumerable highly complex 
and even highly adaptable mechanisms of behavior in which learning 
plays no part at all, but there are none which can dispense with the func¬ 
tion of innumerable mechanisms exploiting instant information. Even 
while a mallard drake performs the grunt whistle, the most rigid, fixed- 
intensity motor pattern I can think of at the moment, his gravity recep¬ 
tors continué to function and his orienting mechanisms keep his spatial 
relation to the courted female constant. 

The functions under discussion are not experience, but they are the 
prerequisite for making experience possible, thus strictly conforming to 
Immanuel Kant's definition of what is "a priori." 

2. The Regulating Cycle or Homeostasis 

The simplest and probably the phylogenetically most primitive form for 
acquiring instant information is the regulating cycle or homeostasis. It 
enables the organism to regain and maintain the equilibrium of its life 
processes after they have deviated in some way or been made to deviate 
from the desirable State. When an animal, under the influence of a lack 
of oxygen, begins to breathe faster, or when, surrounded by abundant 
food, it stops eating for a while, this means that the organism is 
informed, not only about its internal need for certain materials, but also 
about their present availability in its environment—about the condition 

4. Amoeboid Response 


of the "market." The genetically programmed structure of regulating 
cycles makes it possible to maintain, approximately constant, certain 
"valúes of reference" (Sollwert) within the organism. Regulating cycles 
are the most common type of mechanisms exploiting instant informa- 
tion. They range from rather simple Chemical processes to the most 
highly differentiated organizations of neural and sensory organs—as 
will be discussed in Section 7 of this chapter. 

3. Excitability 

Excitability in general, not the action-specific excitability discussed in 
Two/III/3, is usually defined as the readiness of living matter to respond 
to a stimulus. However, neither the concept of "response" ñor that of a 
"stimulus" can be defined exactly. Basically, any Chemical change within 
the organism caused by an external influence is a "response," and any 
such influence can be thought of as a "stimulus." Conventionally, phys- 
iologists think of a stimulus as an external agent causing the organism to 
react by movement or by secretion, and the conceptual división of stim¬ 
ulus and response is derived from the functional división of labor 
between the nervous system, which receives stimuli, and the muscles or 
glands, which respond to them. 

But this duality of organs should not enter into the definition of excit¬ 
ability because not only in all protozoans, but in very many of the lower 
metazoa as well, it is the same cell which receives the stimulus and 
responds to it. In sponges there are the contractile cells surrounding the 
outflow apperture. What ought to be included as part of the definition of 
excitability is the energetic relationship between stimulus and response: 
the stimulus supplies incomparably less energy than is liberated through 
the response. The term "trigger causality" contains an elucidating 
description of the process. 

The only response to stimuli in many sessile metazoa is a contraction 
of the whole body that results in a minimizing of the vulnerable surface 
and, in some cases, a thickening of the protective outer layers. An anal- 
ogous behavior is found in some protozoa, as in amoebas. There seems 
to be no unicellular animal or plant known in which locomotion is not 
correlated to excitability. Both together are programmed in such a way 
as to bring the organism into the best possible environment and to main¬ 
tain it there. 

4. Amoeboid Response 

The most simple and probably the most ancient movement is the plasma 
flow as it is known in amoeboid cells. Rather surprisingly, it shows a 
faculty which otherwise is found only in the movements of much higher 


VI. Mechanisms Exploiting Instant Information 

animáis: the faculty of orientation within the three dimensions of space. 
Orientedness in all possible directions is possible to the "naked" amoe- 
bas for the simple reason that, "as yet," they do not possess head or tail, 
ñor any differentiated parts of body surface and can therefore extend a 
"pseudopod," that is, a hernia-like extrusión of their ectoplasm, in any 
desired direction in space. 

Formerly it was believed that these extrusions were caused by changes 
in the surface tensión of the protoplasm, which was visualized as being 
liquid in all of its layers. In reality, this process is brought about by the 
capacity of the ectoplasm to change, through the influence of a stimulus, 
from the jellied State to the liquid State and back again—something I 
asserted a long time ago, purely on the basis of observation. This has 
since been proved by L. V. Heilbrunn (1958). Normally, that is to say 
always, except at the rrioment of ingesting some object such as food, the 
surface is covered by a thin film of jellied ectoplasm. On the impinging 
of a stimulus eliciting a negative response, this layer is very rapidly 
thickened and, because the protoplasm in the State of gel occupies a 
slightly smaller space than in the State of sol, this thickening results in 
a contraction of the surface, which in turn exerts a pressure on the inte¬ 
rior of the cell—that much of the surface tensión theory is correct. Con- 
versely, the positive response to a stimulus consists primarily in a thin- 
ning of the jelled ectoplasm layer, which causes the body to bulge in the 
direction of the stimulus. At lower intensities of response this stimula- 
tion merely determines the direction in which the animal will move. The 
motion itself has other and not yet analyzed causes. In the most intense 
form of positive response, as in that of taking food, a break in the jelled 
skin permits the liquid endoplasm to spout like a geyser, quickly envel- 
oping the object. The liquid endoplasm then jells on the outside, remain- 
ing liquid where in contact with the object, and in this way the object is 
received into the interior of the cell. 

The information telling the amoeba which objects to ingest and which 
to reject is contained exclusively in the objects' Chemical properties that 
cause the ectoplasm to dissolve. The selectivity of this archaic releasing 
mechanism is none too great. The negative response is elicited by an 
even greater range of stimuli, among which tactile and thermic ones play 
an important role. A needle prick is responded to in much the same way 
as a locally applied heat stimulus. 

In its natural environment, in other words, in a culture in which it can 
live and propágate, the behavior of an amoeba appears so adaptable and 
plástic that the observer tends to forget that all he is seeing are but vari- 
ations in intensity of one and the same process. The animalcule "plac- 
idly" wanders about "searching" for food; it moves hurriedly to avoid an 
unfavorable situation, it escapes from damaging stimuli by "fleeing as in 
panic"; it rushes "hungrily" at prey and engulfs these most "greedily." 
H. S. Jennings, who knows protozoa better than anybody else, says that 
were an amoeba as large as a dog, one could not hesitate ascribing to it 

6. Phobic Response 


the faculty of subjective experience. Yet it is only the change from sol to 
gel and back again that causes the whole gamut of highly teleonomic 
responses whose adaptiveness rests on the selectivity of response to spe- 
cific stimulus situations. 

5. Kinesis 

All unicellular creatures that, in the interest of swimming speed, have 
evolved an elongated and more or less streamlined form as well as a pre- 
ferred direction of locomotion and, therewith, a front end and a rear end 
to their bodies, stand in need of steering mechanisms which guide the 
fast, though rigid, form teleonomically within the three dimensions of 
space. Most freely moving metazoa are faced with the same problem. 
There are but few among them that are radially symmetric and can move, 
if only on one plañe, in any direction they choose. 

One of the most primitive behavior mechanisms, which accomplishes 
the task of keeping swiftly moving organisms as much as possible out of 
unfavorable and within favorable conditions, does so by the simple 
means of slowing down their movements when traversing favorable 
environments and speeding them up while going through unfavorable 
ones. Many of the flagellates and ciliates are more or less continuously 
moving through the water, much the way pelagic fish, such as mackerels, 
do. The behavior just described results in their being found in greatest 
density at favorable locations. By an analogous process, cars are crowded 
in an undesirable density along bad stretches of road that enforce slow 
driving. This mechanism of spatial orientation, according to G. S. Fraenkl 
and D. S. Gunn, is termed kinesis (1961). 

Most organisms do not move in an absolutely straight line; the proto- 
zoans mentioned above usually swim along the line of a screw—regu- 
larly and wrongly described as a spiral. When orienting to favorable 
localities, the eífect of kinesis can be improved by increasing the angle 
of the random deviations from the straight line, and these are inherent 
to locomotion in any case. By this means, the organism is kept in the 
desirable environment longer and is made to exploit an increased part of 
its area. This process, termed klinokinesis by Fraenkl and Gunn, is found 
not only in swimming protozoa but also in higher crustácea, in grazing 
mammals, and in man when, for example, he is searching for 

6. Phobic Response 

Phobos is the Greek word for fear, and the ñame of the orienting mech¬ 
anism here under discussion is derived from its function, which in most 
cases causes the animal to flee from unfavorable stimulus situations. 


VI. Mechanisms Exploiting Instant Information 

However, phobic response is by no means the only process achieving 
this teleonomic effect, ñor is escape the only function of all phobic 
responses. The essential character which defines phobic response is 
found in the fact that the angle at which the animal changes course is not 
dependent on the direction from which the impinging stimulus comes. A 
paramecium which, while swimming forward, moves into rapidly dete- 
riorating environmental conditions, first reverses the direction in which 
its cilia are beating so that it retraces, for a certain stretch, the path along 
which it has come. Then it stops for a time, moving neither forward ñor 
backward, by making the cilia on one side of its body beat in reverse 
while on the other side they resume their forward beating. As an effect 
of this, the animal's front end swings around in a circle, the longitudinal 
axis of its body describing the mantle of a cone. After some time, the 
duration of which depends on the strength and not on the direction of 
the stimulus, the cilia are again switched to full speed ahead and the 
animalcule resumes its travel in a direction which depends exclusively 
on the position of its body at the moment of the switch. 

This direction is chosen at random, in the sense that it has nothing 
whatever to do with the direction from which the stimulus has come. It 
happens often enough that the circle described by the front end mea- 
sures exactly 360° and that the animal resumes its forward motion in the 
former direction, swimming a second time straight into the releasing 
stimulation. If this is not a solid obstacle but a gradient of temperature 
or of a Chemical solution, it often happens that the paramecium resumes 
swimming forward in a direction still less favorable than the former one, 
so that it enters into an even steeper gradient than before. In this case, 
it simply repeats the response until its new direction avoids the stimulus. 

The phobic response furnishes the organism with a much greater 
amount of information than does kinesis. It causes the organism not only 
to shorten its stay in an area having unfavorable conditions, but to avoid 
the area altogether. It conveys to the animal not only the message that 
environmental circumstances are bad, but also in which direction they 
lie. It does not, however, give any report on an improvement in environ¬ 
mental conditions. No reaction is elicited when the paramecium enters 
optimal circumstances; it continúes its straight, though "screwy" path 
until it leaves the favorable area on the other side. Having done this, the 
animal gives the phobic response. A long time ago, when his mastery of 
English was deficient, Otto Koehler once compared this with human 
behavior, saying: "That is just like man; you give him more money, he 
says not anything; you take away money, he cry awful." In Figure 22, a 
diagram by H. S. Jennings (1910) illustrates the phobic response. 

It must be stated that a paramecium possesses other orientation mech¬ 
anisms in addition to the phobic response. The latter is most often seen 
when the animal is observed within the very narrow space of a "hanging 
drop." Under comparably confined spatial conditions a bird, too, would 

7. Topical Response or Taxis 


Figure 22. Schematic representation of single phases of the escape reaction of a 
paramecium which encounters a solid object or some other stimulus in its path. 
(From Jennings's adaptation of Hempelmann, Tierpsychologie.) 

fail to show the more differentiated orientation responses otherwise at 
its disposal. Waltraut Rose has shown incontrovertibly that a parame¬ 
cium possesses orienting mechanisms of the type described in the next 
paragraph (1964). In the larger space of a cuvette, in which no sharply 
angled changes of course are forced upon it, a paramecium avoids unfa- 
vorable situations and remains within favorable environments by means 
of topical responses. One single individual, observed for more than three 
hours, traversed the optimal area in the middle of the cuvette many times 
and turned back into it without once showing a phobic response. The 
observation time was limited not by the activites of the paramecium but 
because Waltraut Rose fainted at her post. 

7. Topical Response or Taxis 

In his classic book on the orientation of animáis in space, Alfred Kühn 
(1919) has described a number of orientation mechanisms which can be 
defined by their common effect of turning the animal in a direction 
directly determined by that of the impinging stimulus. The angle of the 
turn is directly determined by the one between the longitudinal axis of 
the animal's body and the direction of the arriving stimulus. The ñames 
Kühn has chosen for the different types of orienting mechanisms are 
derived primarily from their most common function, just as that of the 
phobic response, which he also called "phobotaxis" in the first editions 
of his book; later he reserved the term "taxis" for topical responses. Some 
of the terms he chose also suggested physiological explanations, working 
hypotheses that have since proved incorrect. Because of this the terms 
should be abandoned. 

The simplest topical response known is termed tropotaxis and consists, 
according to Kühn's definition, of the animal's turning sideways until 
two symmetrically situated sense organs report equal stimulation. This 
was assumed by Kühn to be the case in the orientation to gravity 


VI. Mechanisms Exploiting Instant Information 

observed in vertebrates and crustácea. Among the former, Erich von 
Holst's analysis revealed much more complicated processes, including 
the spontaneous generation of excitation, a discussion of which would 
lead us too far afield. One orientation mechanism for which the theory 
of an equilibrium of stimulation of two symmetrical organs holds abso- 
lutely true is the "rheotaxis" of some flatworms. One flatworm, Planaria, 
reacts with a "positive tropotaxis" to water currents carrying the scent of 
food. Planaria turns against the current until both sides of its tricornered 
head are stimulated equally; then it proceeds upstream. If the head is 
stimulated symmetrically by two jets of water, the worm works its way 
through the resultant rush between the two currents. 

All topical responses are vastly superior to the phobic in regard to the 
amount of information procured. While phobic responses tell the animal 
only in which direction not to proceed, the topical response informs the 
organism unequivocally, which one to choose among all the directions 
possible in space. 

Here is not the place to summarize the results which research on ori¬ 
entation has brought to light since the noted writings of Alfred Kühn 
were published. The part played by complicated regulating processes 
within the sensorineural apparatus of orientation can well be shown in 
the light-compass orientation of insects, which Kühn defined as a meno- 
taxis (from the Greek meno —"I remain"), those orienting mechanisms 
through which a constant direction is maintained by steering a course at 
a constant angle to the incidence of the stimulus. This is achieved by 
feeding the output of the regulating cycle back into its input, thus ena- 
bling the organism to sustain constant valúes of reference (Sollwert). This 
principie of negative feedback is essential to regulating cycles. 

An analysis of the regulating cycles functioning according to this prin¬ 
cipie is complicated by the fact that the choice of a given valué of refer¬ 
ence is subject to higher levels within the central nervous system which 
can, at any time, "command" a new valué (see Two/IV/5). An insect 
steering by the angle of light incidence may, of course, arbitrarily change 
to another course; in fact, it does so every few moments when, for 
instance, it is following a learned path habit. However, it could not stay 
on any course for even one second if it were not for the light-compass 
orientation. That the latter is indispensable becomes apparent when the 
prerequisites for its functioning are lacking. One of these prerequisites 
is that the source of light be at a very great distance so that its rays are 
practically parallel to one another. As we know, lights at a lesser distance 
are a danger to insects. If insects choose for their light-compass a valué 
of reference having an angle less than 90°, they necessarily describe a 
spiral ending at the light itself, just as we see them doing at every Street 
lamp. The lucky ones that have chosen an angle greater than 90° are not 
seen by us at all because they have disappeared into the darkness. 

8. Telotaxis or "Fixating' 


8. Telotaxis or "Fixating" 

Kühn (1919) defines telotaxis as those orienting mechanisms which are 
akin to menotaxis insofar as they function to keep a certain stimulus con- 
stantly at the same place on a certain receptor organ, in most cases the 
eye. The animal turns its eye, its head or its whole body in the direction 
of the stimulus. If, colloquially, we say that we "fixate" on an object, we 
mean that we turn our eyes so as to get the image on the fovea centralis, 
the place of sharpest visión on our retina. There the image is retained, 
even if the object moves quite fast. Our eyes and our head and, if nec- 
essary, our whole body move in such a way as to keep the image at that 
place on the retina. Although it remains stationary there, we perceive 
the movement clearly. Amazingly, our perception reports a smooth and 
continuous movement of the bird we see flying across the sky, notwith- 
standing the disparity of the receptors reporting it. Only during the very 
first moments, just when the bird "catches our eye," is there a succession 
of retinal elements reporting the movement of the bird's image. The next 
instant, when our eyes have fixated on the bird and are following its 
movement across the sky, the message informing us about those move- 
ments stems from the motor organization, studied extensively by Erich 
von Holst and Horst Mittelstaedt, that keeps the eyes fixated on the 
object. In particular, it is the "efference copy" of the motor impulses 
going to the eye muscles that conveys the information about the chang- 
ing direction of the eyes and, therewith, the direction in which the object 
pursued by the eyes is moving. When finally we turn our heads to keep 
that bird in sight, the task of reporting its movement is passed on to the 
neck muscles. Marvelous enough is the fact that all of these messages are 
integrated into one continuous perception of the bird's flight. 

For the catching of prey by the "praying" mantis, a subject particularly 
suitable for this kind of research, Mittelstaedt has demonstrated the func¬ 
tion of the regulating cycles that enable this curious insect to capture 
prey by a stroke of its clawlike front legs (1957). As far as I know, this is 
the most clearly analyzed example of a telotaxis. The mantis fixates on 
prey by turning its head towards an intended victim and, by doing this, 
follows the prey's movements. This, in an insect, is so very unusual as to 
be extremely striking when first observed. If the mantis must creep for- 
ward towards its prey, it turns not only its head in the direction of the 
prey, but also its prothorax to which the catching claws are affixed. 
Finally, if it has rather far to go, it turns its whole body. On the other 
hand, when the prey is so near that no locomotion is needed, the mantis 
can capture it by moving the claws sideways, in a direction not lying in 
the plañe of symmetry of the prothorax. The information underlying the 
insect's very exact aim must be supplied by messages coming first from 


VI. Mechanisms Exploiting Instant Information 

the eyes and then from afferent processes which teii the mantis the exact 
position of its own head at the moment of striking. 

As Mitteistaedt has shown, this positionai information is furnished by 
the "neck organs," cushions of sensory hairs situated on both sides of the 
joint connecting neck and head. At each turn of the iatter, these sensory 
hairs are bent and report the degree of bending. When Mitteistaedt cut 
the nerve ieading from the ieft neck organ, the mantis missed the prey 
by striking too far to the right. If he disconnected both organs, the mantis 
couid hit its prey, but oniy when it happened to sit exactiy aiong the 
median plañe of the prothorax. If it was situated to the right of the plañe 
of symmetry of the prothorax, the mantis wouid miss on the Ieft side, 
and vice versa. So the function of the neck organs is to aim. At first it 
seemed plausible to assume that orientation of the striking movement 
was determined by a summation of the optical stimuli coming from the 
eyes and the proprioceptor reports furnished by the neck organs. If this 
hypothesis of a simple addition of optical and proprioceptor stimulation 
were correct, it should not have made any difference for aiming if the 
head were immobilized in a sideways position by a droplet of glue. Since 
this fixed position of the head wouid be recorded by the neck organs, the 
calculation should still give a correct reading. A mantis with this kind of 
immobilized neck can still fixate by turning its prothorax, or even its 
whole body, in such a way that the prey lies along the plañe of symmetry 
of the head. However, having done the turning, the mantis still misses 
the prey by striking on the side opposite to that in which the head is 
turned. As Mitteistaedt says, it acts as if it did not know that its head is 

To be concluded is that the aiming mechanism of the mantis can 
achieve a correct report on the position of its head only when the head 
is able to move freely. The neck organs report the position of the head 
relative to the prothorax, but this report is not fed directly into the mech¬ 
anism aiming the strike of the anterior legs, but into the motor apparatus 
of the neck muscles. Figure 23 is a diagram of the processes involved in 
the aiming done by the mantis. Mitteistaedt observes that the mantis fix- 
ating on a fly sitting to its right aims its claws according to the amount 
of innervation expended to effect that position of the head—or to put it 
in an anthropomorphic way, the mantis directs its strike in that direction 
it believes its head is aiming. The information concerning the actual 
position of the head contained in the reports of the neck organs is not 
passed directly to the apparatus of localization; instead, this information 
is fed into a lower motoric center whose task it is to make the normal 
position (the "zero" position) of the head independent of deviation 
caused by any drag on the neck musculature. This can be proved exper- 
imentally: if minute weights are attached laterally to the head so as to 
twist it sideways, a considerable amount of torque can be applied without 
any decline in aiming accuracy. 

9. Temporal Orientation 


Figure 23. Functional diagram of the mechanism underlying localization in man- 
tids. The hypothesis is developed by steps from (a) via ( b ) to (c). (a) Optic feed- 
back loop only. As indicated by the arrows, information flows from the optic unit 
(amplification factor: A (opt) ) to the neck motor unit (amplification factor: A (neck) ) 
and again to the optic unit. ( b ) Optic and proprioceptive feedback loops. The neck 
motor unit is controlled by the difference between the optic (</> c ) and the propri¬ 
oceptive (<5 C ) center messages. ( c ) Complete hypothesis. A (stroke) amplification factor 
of the central unit which determines the direction of the stroke (k). For full expla¬ 
naron, see text. (Mittelstaedt, H.: "Prey Capture in Mantids.") 

9. Temporal Orientation 

Like everything else in the universe, behavior proceeds within space and 
time. Behavior must occur not only in the right place and with the right 
orientation, but also at the right moment. Quite generally, releasing 
mechanisms, whether innate or acquired, are programmed so as to set off 
adequate behavior at the teleonomically correct time. In this respect 
IRMs can be regarded as mechanisms exploiting instant information. 


VI. Mechanisms Exploiting Instant Information 

Time of Day 

Figure 24. Periodic behavior of an experimental subject (human) in a bunker that 
was completely isolated from the outside world. The dates on the left indicate the 
beginning of a waking period (heavy horizontal bars). (Aschoff and Wever, 

Besides these "time givers" relying on outside information, most 
organisms possess "internal docks" that, independently of any outside 
information, are working in cióse synchronization with the great cosmic 
cyclical processes. They are able to tell the organism, with greater or 
lesser exactitude, "when the bell has tolled," whether it is day or night, 
summer or winter. Many marine organisms are, furthermore, informed 
about the runs of the tides. 

C. Pittendrigh (1958) and J. Aschoff (1965) concentrated most of their 
temporal investigations on the "circadian" rhythms, e.g., those which 
correspond to the turning of the earth on its axis. Internal docks attuned 
to these rhythms were found to keep time much less exactly than any 
tolerably precise, man-made chronometer. There are some internal docks 
which are "fast" or "slow" by a certain amount of time, and this amount 
remains surprisingly constant over considerable periods, as shown in 
Figure 24. Within its natural surroundings, the organism corrects the 
error of its dock every day using external "time givers." The necessity of 
this correction is so widely spread that Pittendrigh and Aschoff tend to 
suspect the existence of unrecognized time givers whenever one of these 
internal docks appears to be absolutely exact. No organism is known to 
be capable of exploiting the information received from external time giv- 

10. Navigation by Sextant and Chronometer 


ers in order to speed up or to slow down the running of its own dock. 
It cannot resort to any kind of correction analogous to that which the 
clock-maker would perform by adjusting the length of the pendulum, 
thus making the dock run faster or slower. All it can do is what a person 
who is not an expert does to his watch or dock, that is, advance or set 
back the hands each day. 

There are internal docks with "hands" that are easily adjusted and oth- 
ers that are very resistant to such "settings." This is true not only for 
circadian rhythms but also for those covering longer periods of time. The 
annual rhythms of two birds belonging to the same order of Anatidae 
represent both extremes. Cañada geese, Branta canadensis, transplanted to 
New Zealand, have been reliably reported to be nesting and breeding 
during their first Southern spring; black swans, Cygnus atratus, natives of 
Australia, when brought to Europe persist for years and, in some cases, 
for generations in breeding during the northern autumn. Among 
humans, the circadian dock can be very "soft" and adjustable in some 
people who, even when travelling halfway round the globe, still find it 
easy to adapt; it can prove to be extremely "hard" in some less fortúnate 
people who suffer for many days from such a need to reset their docks. 
Aschoff and his co-workers have shown that time changes of this type 
are not always harmless to the organism: in some insect species the nor¬ 
mal duration of life is considerably shortened if they are forced to 
undergo repeated "resettings" of their circadian docks. 

10. Navigation by Sextant and Chronometer 

As anyone who has read South Seas stories knows, a navigator whose 
chronometer has gone wrong can "fetch" an island by "sailing down its 
latitude," for example, by using the polar star as a point of reference. In 
order to steer a straight course by the sun, however, its movements have 
to be taken into consideration and to do this a chronometer is indispens¬ 
able. Furthermore, from the chronometer readings and observations of 
the sun's movements ("shooting the sun"), longitude can be computed. 
Some social insects which always return to their nests after an absence of 
very short duration can afford to neglect the sun's movements. They nav- 
igate, on the way out, by means of the light-compass orientation 
described in Section 7 of this chapter, and return by simply reversing the 
process by 180°. If, as shown in Figure 25, one covers with a cup or bowl 
a worker of the ant species Lasius niger on its way out and keeps it in 
darkness for a period of time, it will, on being liberated, return home 
deviating from the correct course at an angle corresponding to the extent 
which the sun has travelled during the period of imprisonment. 

For trips of longer duration, obviously this method of orienting could 


VI. Mechanisms Exploiting Instant Information 

0 0,5 1 m 

Figure 25. Time experiment with Lasius niger, journey back as a reverse of the 
journey out. The sun-compass orientation alone, which does not take into account 
the sun's movement, leads to a corresponding aberration. (Brun, adapted from 
Kühn, in Hempelmann, Tierpsychologie.) 

not be used. Many phyla of animáis, vertebrates such as fish and birds, 
as well as some arthropods have, independently of one another, evolved 
highly differentiated computing mechanisms enabling them to inelude 
the sun's azimuth in their computations and allowing them to steer 
straight courses even on long journeys. 

A starling hand-reared from the egg by Klaus Hoffmann (1960) which 
had never seen daylight, let alone the sun, was kept in a dark room illu- 
minated by an artificial sun. In order to make the bird "accept" this light 
as the sun, it proved necessary to simúlate risings and settings by moving 
the light along the vertical. The bird was kept in a radially symmetrical 
cage and was trained to look for its food in trays also arranged symmet- 
rically all around the enclosure. For a pilot study, the bird was taught to 
feed at six o'clock in the morning from the tray lying directly under the 
artificial sun which had just "risen." After completing this conditioning, 
the starling was tried at noon, and it promptly went to that tray situated 
ninety degrees to the left of the sun now standing at the highest point 
of its up and down movements. In other words, the bird had learned to 
search for food "east of," and not "below the light source." Within a dark 

11. Taxis and the Fixed Motor Pattern 


room, the starling could have been used as a dock; out in the open, it 
could have been used as a compass. 

11. Taxis and the Fixed Motor Pattern 

As has already been explained in Two/VI/1, it is a fundamental error to 
believe that learning processes take part in all the action patterns of 
higher animáis. On the other hand, the mechanisms exploiting instant 
information discussed in this chapter actually do participate in practi- 
cally all behavior. As was shown in Section 2, regulating cycles are prac- 
tically omnipresent and orienting mechanisms function wherever loco- 
motion occurs. In organisms sufficiently evolved to possess a central 
nervous system and centrally coordinated motor patterns, it is all but 
impossible to find examples of behavior in which topical responses do 
not play a role. At least those responses maintaining equilibrium are 
always present. I can think of only two behavior patterns in which ori- 
entation to gravity might be said to disengage. In the spawning of the 
fighting fish, Betta splendens, both partners seem to "faint," drifting in 
the water for a few seconds devoid of all orientation; the male rabbit does 
something analogous during copulation—he falls sideways off the 

Terrestrial vertebrates which move their legs in absolute coordination 
for locomotion (as was discussed in Two/I/13, 14) add or subtract, by 
means of the orienting mechanisms, a certain amount of motor impulse 
at every step, just enough to adapt the movements to the irregularities of 
the terrain. In Three/IV/7, 8, I elabórate on these processes. The whole 
system of orientation mechanisms superimposed on centrally coordi¬ 
nated movements has been called by Erich von Holst "the mantle of 

As already mentioned, the early literature often ineludes the processes 
of locomotion as part of the concept of taxis. Thus, "positive phototaxis" 
does not only mean that the insect turns toward the light, but also that 
it flies towards it; "negative geotaxis" implies that the organism strives 
to move upwards, and so on. The very same reasons that impelled us 
(Two/I/1) to divide Heinroth's concept of the species-characteristic drive 
action into its component parts make it advisable to formúlate the con- 
ception of turning in space as something sepárate from the conception of 
locomotion. As in a ship, the apparatus for turning the rudder to port or 
starboard is expected to be different from, and rather independent of, the 
apparatus for switching the motor power on and off, and certainly dif¬ 
ferent and sepárate from the machinery of the engines turning the pro- 
pellers. The "positive Europotaxis" of an ocean liner is brought about by 
the functions of the turbines, of the steering apparatus, and, last but not 
least, by the functions of the navigator. This example delineates rather 


VI. Mechanisms Exploiting Instant Information 

well the limited influence the higher levels exert upon the lower ones. 
The highest levels of the nervous system cannot change the basic func- 
tions of fixed motor patterns any more than a captain can influence the 
possible movements of his ship's engines. He can command only their 
intensity and signal slow ahead, full speed ahead, or, when necessary, 

The multiplicity of the constantly participating orientation responses 
tends to conceal somewhat the rigidity of fixed motor patterns. In this 
respect they can have an effect similar to the changes in intensity already 
discussed in Two/I/3. It is certainly not a coincidence that Whitman and 
Heinroth íirst discovered fixed motor patterns through examples in 
which both of these effects were absent, that is, in those activities barely 
overlaid by any orientation responses and, furthermore, with fixed 

Using the egg-rolling behavior of greylag geese, Tinbergen and I 
attempted to analyze the way in which a fixed motor pattern cooperates 
with an orienting mechanism (1938). Egg rolling is conducive to analysis 
because it permits an experimental separation of the two components. 
The behavior is released in an incubating goose when it sees an object 
situated outside and near the nest and having a number of releasing 
characteristics. The object must have a smooth outline devoid of all pro- 
tuberances, and it must have a hard surface. Its form is irrelevant; 
wooden cubes are treated as if they were eggs, and the size can vary from 
a few cubic centimeters to the máximum that can be encompassed by the 
goose's neck. The fixed motor pattern used in the action consists of a 
stretching forward of the neck, bending the head downward so as to 
touch the egg with the underside of the bilí, and then rolling it toward 
the nest by means of a slow bending of the neck. Concomitant compen- 
satory movements of head and bilí to each side keep the egg in balance 
and prevent its deviating from the intended path. The fixed motor pat¬ 
tern can be isolated by deftly snatching away the egg after the movement 
has been released. The movement then continúes to run smoothly all the 
way through to the nest cup, staying strictly within the median, that is, 
along the bird's plañe of symmetry. Once the movement has been 
released, it can only run its way through to the end and can be changed 
neither in its coordination, ñor in its strength. If one offers the goose an 
object much too large, such as a huge cardboard easter egg, the move¬ 
ment literally "jams"; the goose proves unable to move the object in any 
other than the prescribed way—for instance by walking backward. If the 
object is not heavy enough, it is lifted off the ground; if it is too heavy 
by even only a slight amount, the movement fails to budge it. This is 
remarkable because a goose's neck is capable of producing a prodigious 
amount of power, for instance, enough to pulí a table cloth loaded with 
a complete tea set off a table or, in a more teleonomic way, to tear heavily 
rooted plants out of the bottom of a pond. However, the power at the 

12. Taxis and Insight 


disposal of the fixed motor pattern is strictly measured to serve its single 

As can easily be demonstrated, the movements to each side, which dur- 
ing the whole procedure keep the egg balanced on the underside of the 
bilí, are elicited by tactile stimuli emanating from the object. Whenever 
the egg deviates to one side, the bilí immediately follows it and guides 
it back into the right direction. It is possible to make the egg "run on 
rails" during the rolling process by arranging a bundle of reeds obliquely 
across its path. Then the movement tries to overtake the egg in order to 
correct the "wrong" direction and sometimes succeeds at the moment 
when the pressure of the bilí acts at a right angle to the obstacle. If the 
goose "rolls" a square object that facilitates the establishment of a stable 
contact with both branches of the mandible and thus not diverging from 
the straight line either to the right or to the left, the balancing move¬ 
ments cease altogether and the fixed pattern alone predominates, just as 
it does when the object being rolled is removed altogether. 

12. Taxis and Insight 

Conventionally, insight behavior is defined by exclusión; the organism 
is acting through insight, it is said, whenever an instant solution is found 
to a problem for which the organism possesses neither a phylogeneti- 
cally programmed answer ñor one acquired by learning during the 
course of its individual life. 

A better definition of insight behavior can be given. When a fish per- 
ceives a desired object behind an obstacle, for instance a chironomous 
larva behind a finely branched bush of myriophyllum, through which it 
cannot swim although it can see the larva, even the most stupid goldfish 
will swim around the bush to reach the tidbit. This detour can be 
explained as the result of a positive telotaxis directed toward the prey 
and of a negative thigmotaxis causing the fish to avoid contact with the 

Between this simplest discovery of a detour and the most complex 
functions of insight in animáis, such as those investigated by Wolfgang 
Kohler in chimpanzees (1963), there exists a continuous gradation of 
intermediates bridging the gap. Methodos is the Greek word for "detour," 
and the expressions which our natural language has found to describe 
our highest intellectual achievements bear the ineradicable marks of 
their provenience from the mechanisms of orienting behavior in space. 
We gain "insight" into a "maze" of circumstances that can be very "com¬ 
plex," just as an ape does into a tangle of intertwined branches through 
which it has to wend its way. The "concept" we form of a thing cannot 
become "clear" before we have thoroughly "grasped" it. In the latter 
expression the priority of tactile perceptions becomes apparent. Through 


VI. Mechanisms Exploiting Instant Information 

insight we understand only that which can be represented or "visual¬ 
ized" in the spatial model our perception projects in our central nervous 
system—and in our consciousness. W. Porzig knew this more than a 
quarter of a century ago (1950). He wrote then: 

Our language translates everything that cannot be visualized into spatial 
concepts. This is not done just by our own language, or by a certain group 
of languages, but by all of them, without exception. It is a property (an 
"invariant") inherent to human language as such. Temporal circumstances 
are expressed spatially: "before" or "after" Christmas, two years have 
"passed," a "space of time" of several minutes has "elapsed," etc. Concern- 
ing subjective processes we speak not only of the "innermost soul," of a 
"threshold" below which our consciousness is not affected, of the "sub"-con- 
scious, of "layers" of consciousness, and so on. Space serves as a model for 
everything that cannot be visualized. "Besides" his professional work, he 
occupies himself with poetry, "between times" he did something else, by 
this act he "covered" his real intention—it is redundant to add further 
examples which can be found in any number of samples of written or spo- 
ken language. The phenomenon under discussion derives its importance 
from its wide distribution and from the part it has taken in the history of 
language. It is apparent in the use of prepositions which originally have a 
spatial connotation and equally in that of adjectives and verbs. (Translated 
from the Germán) 

I once explained all of this in a lecture and, afterwards, one of the dis- 
cussants asked a Chinese student who was conversant in ancient Chinese 
to write on the blackboard "before Christmas" and "after Christmas" and 
"an oak is growing in front of the house and a fir behind it." The ideo- 
grams for temporal and spatial prepositions proved identical. 

I want to add that these considerations of linguistic phenomena are in 
full agreement with the expectations of anyone who has examined ori- 
enting mechanisms. All orientation, including our own, is always con¬ 
cerned with space and time. We are quite incapable of perceiving or even 
of visualizing space without simultaneously visualizing something hap- 
pening, in other words, movement in space, ñor are we capable of visu¬ 
alizing movement without a space within which it can occur. 

We define as "directed by insight" any behavior the teleonomy of 
which is based on the function of mechanisms exploiting instant infor- 
mation. It proves quite impossible to draw a sharp distinction between 
the simplest detour of that goldfish and the most complex insight- 
directed behavior of Wolfgang Kohler's chimpanzees, ñor can we give a 
definition capable of setting human insight apart, or the "methods" gov- 
erned by insight, as something essentially different. The fact that learn- 
ing always participates in the more complicated processes of insight can¬ 
not be used in a definition, since sometimes it is just as involved in quite 
simple Solutions to detour problems. Even for these Solutions the reports 

12. Taxis and Insight 


of one mechanism exploiting instant information must be remembered 
long enough to be integrated with the messages of another or several 
others, so that a response which takes all of them into account can be 

There is a special type of learning, often termed "insight learning," 
the most primitive form of which is found in organisms whose processes 
of orienting in space do not take place simultaneously with the move- 
ments they are directing—as they do in the egg-rolling movements per- 
formed by geese. Instead they achieve their effect before the movement 
is set off as, for example, in taking aim before firing a shot. The processes 
orienting the locomotion of fish furnish examples of both types. Many 
fish, particularly those living in open water, orient by the parallactic shift 
of the images projected onto their retinas. We can do the same while 
riding on a train at night, when we can see nothing outside except the 
lights of the lamp posts. Our perception is perfectly capable of calculat- 
ing not only the distance from us each lamp post is, but also the speed of 
the train's movement which makes the posts shift on our retinas. This 
kind of spatial orientation can obviously function only as long as the 
organism is in motion. A goldfish can come to a stop at a place where an 
obstacle is situated directly in front of it. Though it must have received 
information concerning the obstacle while approaching it, the fish, on 
resuming its forward motion, does so directly toward the obstacle, veer- 
ing off only after it has got underway. 

Other fishes, notably those living in an environment full of complex 
spatial structures, gain information about these forms while standing 
still, moving their eyes in all directions, fixating on one object after 
another, thus optically exploring the vicinity. When they move, they 
start at once in the "chosen" direction. As they never collide with an 
obstacle, they obviously must have marked and, for a short span, remem¬ 
bered the direction in which the way was open, where obstacles were 
located and where there was desirable cover. The stickleback is a typical 
example of an animal orienting by looking around and fixating on 
objects successively. Among insects, the praying mantis furnishes the 
only example of an analogous process. Some birds orient by parallactic 
shift, as the goldfish does, but, unlike the latter, they acquire their infor¬ 
mation prior to moving. Typically these are large-eyed birds with 
restricted eye movements. In order to make the images of the objects in 
their vicinity shift on their retinas, some of them, like the red-breasted 
robin, move their heads up and down or left and right; some—as, for 
instance, kestrels, small owls, and others—make both movements. 

A naive human might find this method of gaining information about 
the spatial structure of the surroundings extremely funny, while even a 
more sophisticated observer could not help feeling that organisms ori¬ 
enting by fixating are more "intelligent" than those orienting by paral¬ 
lactic shift. In fact, this impression is not too far wrong, as the telotactic 


VI. Mechanisms Exploiting Instant Information 

or "fixating" method of orientation observed in the "intelligent" stickle- 
back presupposes the participation of a learning process, if only a very 
short-lived one, while the parallactic orientation of the goldfish does not. 

The sequence of events, first orienting and acting only after orienta¬ 
tion has been achieved, as is found in all "fixating" organisms, is encoun- 
tered once again in very high animáis, in monkeys and apes. In these, 
the temporal separation between the processes of gaining insight and of 
acting on it is even more clearly marked. As Wolfgang Kohler most 
graphically described (1963), chimpanzees approach a new problem by 
first "studying" the situation, for example, by exploring it visually as 
extensively as possible. This is most beautifully shown in a film produced 
by the primate laboratory at Sukhumi in the Soviet Union. A young 
orangután is confronted with a problem represented by a banana sus¬ 
pended from the middle of the ceiling and by a wooden box standing in 
a córner. The ape has never met with this particular problem before, but 
he knows that in this lab the solving of problems is, in general, 
rewarded. He obviously has a preformed association between the 
appeted goal and the presence of objects to be manipulated. He looks up 
at the banana, then at the floor directly underneath the banana, and back 
up at the fruit. Subsequently, he shifts his glance to the box, then back 
again to the banana, his eyes moving diagonally across the room. Now 
he seems to know that a) the banana is too high up for him to reach it 
from the floor, and b) that the box, unfortunately, is not under the fruit. 
At this point the orangután goes into a temper tantrum, exactly like that 
of a human child, throwing himself onto his back, kicking his legs, and 
screaming. After a while he quiets down, begins to scratch himself vio- 
lently, in conflict concerning the attraction of the goal and the apparent 
impossibility of reaching it. After this he concentrates anew on the prob¬ 
lem, at first by again looking back and forth between the box and the 
fruit. Then something happens that is really impressive: the ape looks up 
at the banana, down at the spot below it; from there his glance switches 
across the lab to the box and back to the place under the banana and now 
up to it again. In the next moment, with a somersault motivated by puré 
joy, the young orangután rushes across the room to the box, shifts it, and 
gets the banana within seconds. The gaining of insight has required 
many minutes (probably longer than shown in the film); acting on it suc- 
cessfully has taken but a few seconds. 

What this ape has been doing is thinking. By gaining and exploiting 
quite a large amount of instant information, and with the help of some 
insight learning, he has built up within his central nervous system a 
model oí his spatial surroundings and he is acting internally and within 
this visualized representation of space—and he has done so without 
moving anything except his eyes. Acting in visualized spaces precedes 
bodily action and it can, within limits, comprise genuine exploratory 
behavior employing the trial and error method. Some higher mammals. 

12. Taxis and Insight 


primates in particular, are thus able to act within a visualized three- 
dimensional model of their spatial surroundings and solve the problems 
presented by these surroundings. I maintain that all human thought pro- 
cesses are based on the same kind of acting in visualized space. 

The sameness of the principie ruling the simplest functions of orient- 
ing mechanisms as well as the highest achievements of human insight is 
strikingly expressed in the concomitant subjective phenomena. Karl 
Bühler has called attention to the fact that an identical and unmistakable 
subjective phenomenon—termed the "Aha-experience"—always occurs 
at the moment when a State of disorientedness gives place to that of 
being oriented (1922). This is true for the simplest tropotaxis, as I know 
from an unforgettable experience. While I was sound asleep on the deck 
of a motorboat, a jocular friend had what he thought was a good idea: 
that of rolling me overboard. It was a dark night and the Danube is a 
very turbid river so that, when I awoke, I found myself in absolute dark- 
ness. I happened to be suspended exactly upside down and, for a few 
panic-sticken moments, I swam downward. When my statoliths took 
over, telling me unequivocally which way was "upward," I experienced 
an unforgettable "Aha-experience." Phenomenology is—contrary to the 
opinions of some—a legitímate source of knowledge and I am in a posi- 
tion to assert that, on solving a scientific problem, the experience of 
"Aha" is qualitatively identical to the one accompanying the clicking of 
gravity orientation. 

With full justification we feel that our capacity for directing our actions 
by insight represents one of the highest valúes. Our attitude to the ques- 
tion, what is to be regarded as a "higher" and what as a "lower" animal, 
is strongly influenced by this inescapable valué judgment. 

In order to keep the organism constantly oriented and informed about 
the never-ending changes within its environment, and in order to judge 
the priority of incoming insight information, a special control center is 
needed, superimposed on all these orienting mechanisms. The function 
of this center is to zvatch over the interactions of subordínate processes. 
"Watching" implies staying awake and, I suppose, there must be some 
wide-awake, in other words, some spontaneously active processes under- 
lying the function of this highest control in our central nervous system. 
As a mere speculation, I would say that the thread of our consciousness, 
spun without interruption as long as we are awake, is the subjective con¬ 
comitant of this very central function, of a function whose competence 
by far exceeds that of the disinhibiting mechanisms discussed in Two/ 

Chapter VII 

Múltiple Motivation in Behavior 

1. The Rarity of Unmixed Motivation 

For reasons of didactic simplicity I have not, until now, mentioned the 
fact that behavior can be, and in very many instances is, activated by 
more than one motivation. Only in Two/I/13, 14 have I described how, 
on a lower level of integration in which the processes of endogenous 
rhythms investigated by Erich von Holst hold sway, two or more 
impulse-producing rhythms can compete with one another for the mas- 
tery of muscle activity and achieve, by mutual interaction, what Erich 
von Holst has called relative coordination. I have not yet mentioned that 
phenomena comparable to these can be found at the higher level of fixed 
motor patterns. The "classical" oíd ethologists were not aware of this. 
Many years ago my teacher, Julián Huxley, used to express the difference 
between animal and human behavior by means of a parable. He likened 
the animal to a ship commanded by many captains who, however, were 
not on the bridge simultaneously, but had a gentlemen's agreement that 
each of them would cede his command at once if one of the others 
climbed onto the bridge. Huxley likened the human to a ship also com¬ 
manded by many captains, all of whom stay on the bridge continuously, 
each giving his own commands without consideration of any of the oth¬ 
ers. Sometimes the conflict caused by their countermanding commands 
leads to complete chaos, but sometimes they jointly succeed in choosing 
a course which none of them would have arrived at alone. 

In a discussion about progress in Science, Jakob von Uexküll once said: 
"The truth of today is the error of tomorrow," whereupon Otto Koehler 
replied: "No, the truth of today is the special case of tomorrow." In the 
history of Science this assertion can be supported by a great number of 

2. Superposition 


examples among which the most impressive are the developments that 
led from classical physics to quantum theory. At the time Julián Huxley 
used the parable quoted above, it was perfectly legitímate to cling to the 
simplest example available. Discovering the simplest example available 
has played an enormous role in the history of Science; the Mendelian 
laws would not have been discovered—at least not when they were—if 
Gregor Mendel had not hit on monohybrids, the simplest possible case. 
The simplest form of interaction between two behavior patterns is indub- 
itably that of mutual inhibition. This most often functions in the way 
mentioned in Two/IV/5, on the principie of the máximum selecting Sys¬ 
tem, and the particular case Huxley's parable illustrates is the one of 
mutually inhibiting action patterns represented in Tinbergen's diagram 
(see Figure 19b) as those lying on the same level of integration. 

Widespread is the erroneous belief that the cases most frequently cited 
in textbooks are also those which are most often found in nature. In 
nature, behavior activated by a single motivation is found at least as 
rarely as hybrids differing in only one gene. A higher animal in its nat¬ 
ural habitat must always be ready to undertake a great number of differ- 
ent and—as often as not—mutually exclusive actions, and what it finally 
does is almost always a compromise made among several necessities. 

2. Superposition 

The simplest form of interaction between two simultaneously activated 
motivations is "superposition." Even at the level of endogenous 
rhythmic production of impulses, we have already encountered simple 
superposition of excitatory effects in the sense of addition as well as of 
subtraction. At the higher level of fixed motor patterns, additive super¬ 
position is more frequent than subtractive, possibly because subtraction 
tends to be replaced by complete mutual inhibition. Additive superpo¬ 
sition is found even in cases in which the two independent motivations 
activate antagonistic muscles. We all know how humans become quite 
literally "tense" under the influence of conflicting motivations. A con- 
flict between motivations in Anatidae, one demanding a forward exten¬ 
sión of the neck, the other a retraction, which can occur in a goose want- 
ing to eat grain offered in a human hand and not quite daring to do so, 
produces a violent trembling of the neck. 

In most of these cases of simultaneous activation by conflicting moti¬ 
vations, it remains doubtful at what level the conflict actually takes place. 
From the investigations undertaken by Erich von Flolst, we know that 
this can happen at a very low level of automatic rhythms contending for 
supremacy and becoming superimposed upon one another in every mus- 
ele contraction. The "trembling neck" of the goose probably is effected 
by conflicting innervation of the antagonistic muscles, but we cannot be 


VII. Múltiple Motivation in Behavior 

Figure 26. Facial expressions of the dog that result from a superposition of various 
intensities of fighting and flight intentions. a-c: Increasing readiness to flee; a-g: 
increasing aggression and the corresponding superpositions. (Lorenz: On 

certain about this. There are cases in which different motivations rule 
units as large as different locomotor organs, fins for example. In territory 
disputed, certain cichlids (Etroplus maculatus) position themselves oppo- 
site one another, threatening across the border separating their territo- 
ries. As in every threat, aggressive motivation is contending with that 
for escape. Whenever one of the adversaries moves a short distance for- 
ward into enemy territory, it appears as if he were swimming into a cur- 
rent, the speed of which rapidly increases as one proceeds upstream. This 
eífect is produced by the action of the pectoral fins, which are sculling 
in reverse, and doing so more and more intensely the farther the fish 
moves into the other's territory. The tail fin is under the control of 
aggressivity and the pectorals under that of escape, and the observer can- 
not help feeling—ridiculously—that the pectorals are more afraid than 
the tail, because they are nearer to the enemy. The oíd adage about fear- 
ing one's own courage fits this situation perfectly. 

In very many animáis, gestures of threat have originated from such 
superpositions of aggression-motivated and escape-motivated behavior 
patterns. An oíd example illustrating the complicated gradations of 
movements which can arise in this manner is that of the facial expres¬ 
sions of threatening dogs (Figure 26). The intensity of each of the two 
conflicting motivations can actually be calculated by measuring the lin- 

3. Mutual Inhibition and Alternation 


ear extent of the movements of lips and ears. The figure legend serves 
well as an explanation. 

Very complicated eífects are produced by the superposition of the sev- 
eral vocal expressions of the greylag goose. The signáis of alarm, of the 
intention to move, of calling for company, and of distress, as well as some 
utterances the meaning of which is as yet unclear, can be superimposed 
on each other in an indefinite number of ways and proportions. Curi- 
ously enough, all of these combinations are directly intelligible to the 
initiated human observer. In this way a goose can express that it is going 
to move on because it feels lonely, or because it feels that the environ- 
ment has become dangerous, or simply because it does not like the pres- 
ent situation. The fact that some utterances cannot be mixed, for instance, 
the sound of displeasure and the eating sound, or the expressions of love 
and danger signáis, and so on, needs a very special study and gives rise 
to interesting speculations about subjective experience in animáis, par- 
ticularly about the pleasure-displeasure problem. 

Provided they do not demand simultaneous contractions of antagonis¬ 
ta muscles, any two motor patterns can be superimposed on one another. 
As we have seen, superposition is possible even if the motivations of the 
two patterns are apparently opposed to each other, as are aggressivity 
and escape in the goose's tremble-neck, in Etroplus 's territorial fighting, 
and in the dog's threatening. If there is no functional relationship 
between the motor patterns, the possibilities of superposition are vir- 
tually unlimited. One example, given in Figure 27, should suffice. The 
precopulatory movement of the mallard—and of many other ducks— 
consists in a vertical up-and-down pumping of the head, the bilí being 
held horizontal all the while. An important courtship movement of the 
female, the so-called "sicking," consists of alternately facing the male 
and pushing the head in a threatening movement in the direction of a 
"symbolic enemy." Both movements are highly ritualized and their der- 
ivation is well known from comparative studies, but this does not con¬ 
cern us here. The film from which Figure 27 has been drawn is of a pair 
of mallards. The female is shown alternating between the fixed motor 
pattern of pumping and sicking. Once the peaks of the two movements 
happen to come so near to each other that the attraction of the "magnet 
effect" discovered by Erich von Holst sets in, the two patterns "stick" to 
each other in superposition and are thus performed, for a few strokes, 
synchronously and in absolute coordination. 

3. Mutual Inhibition and Alternation 

Only a few instances are known in which the activation of one behavior 
system exeludes absolutely the activation of any other. Fleeing from a 
predator is an example of such an activation having priority. It is obvious 


VII. Múltiple Motivation in Behavior 


Figure 27. a The so-called "inciting movement" of the female mallard is released 
by the "symbolic" drinking and preening movements of the drake. The move¬ 
ment of the duck pointing her head backwards over her shoulder is repeated 
several times, b The "pumping movement" of the head is a species-characteristic 
pre-copulatory movement. c The culmination points of both movement patterns 
come near to coinciding and remain tied to and superimposed on one another in 
the same phase relationship—by virtue of the magnet effect—for the duration of 
several eyeles. 

3. Mutual Inhibition and Alternation 


that running even the slightest bit slower, when traversing a field of 
tempting pasture, would be highly disadvantageous for the haré pursued 
by a couple of greyhounds. Escape motivation of high intensity not only 
suppresses all other instincts; occasionally and in a distinctly dysteleon- 
omic manner it suppresses all the functions of learning and insight as 
well. Not even humans are exempt from the absolutely stultifying effects 
of panic. 

There are few examples of instinctive systems other than that of escape 
having a similarly absolute priority. There are, however, some instances 
in which even escape is suppressed by the influence of another motiva¬ 
tion. In young eels ascending into fresh water, the motivation to migrate 
all but extinguishes escape. So does sex in spawning toads, salmón, and 
some other animáis. 

A special case of suppression of escape by another motivation is what 
H. Hediger has termed the "critical response" (1934). When further 
flight is blocked in some way, for example, when the pursuing predator 
has come too cióse, or when there is an unsurmountable obstacle to 
escaping, the pursued animal quite suddenly turns and attacks the pur- 
suer. "Fighting like a cornered rat" has become proverbial in English. 

Mutual inhibition between two conflicting behavior systems appears 
to occur most frequently among action patterns which, in Tinbergen's 
diagram of hierarchical organization (see Figure 19b), are positioned on 
the same level of integration. This was discussed in Two/VI/3. It would 
obviously be dysteleonomic if activities of this kind were either super- 
imposed or exerted an inhibitive influence on one another. It is of 
obvious advantage if either the one or the other is performed with 
undiminished intensity. This "either-or" presupposes a sort of "filtering" 
mechanism deciding which of the two is to be given priority over the 
other. This mechanism ensures that the motivation which happens to be 
the strongest at the moment is allowed to pass through undiminished, in 
spite of weaker rivalling impulses. As B. Hassenstein has pointed out 
(1973), a comparatively simple interconnection of nervous pathways can 
accomplish this filtering. He says: 

A branching-off of inhibiting effects must lead from every pathway to every 
other, so that whichever path is conducting the strongest flow of impulses 
at a given moment is in a position to block those of all others. There are two 
possibilities: the inhibiting connections leading to neighbouring pathways 
can branch off before or after the comparable connection stemming from 
neighbouring pathways are received. The second of these accomplishes the 
desired effect of letting the strongest motivation pass undiminished while 
inhibiting all others. 

The principie of this neural network, the "lateral backward inhibition" 
(laterale Rückwartsinhibition) was discovered by H. K. Hartline and co- 
workers (1956). 


VII. Múltiple Motivation in Behavior 

The principie of the "máximum selecting system" can ensure that, 
even among more than two competing pathways, the one conducting the 
strongest flow of impulses is let through. It is probable that a great num- 
ber of behavior systems that must be prevented from functioning simul- 
taneously are connected in this manner. As Hassenstein emphasized: 

This functional part of the behavior system ensures that, at any time, an 
activity whose motivation is the strongest suppressess all others. It is not the 
readiness for certain activites that is either suppressed or given free rein, but 
the actual motivation dependent on the internal readiness as well as on the 
momentary stimulus situation impinging on the animal. Therefore, if one 
motivation is damaged, the next strongest is at once able to take over. 

Within the organism, the principie of a double quantification [not the one 
discussed in Two/I/5] and of a máximum selecting system [Extremwertdurch- 
lafi] are combined in the way represented in [Figure 21]. The two-part Sys¬ 
tems modify the tendency toward simultaneous activation of incompatible 
behavior patterns into a temporal sequence. Many motivations increase grad- 
ually, as do those for defecation and micturation. At some moment one of 
them becomes stronger than all others and pushes through completely, to 
subside to zero after having fulfilled its need. The system represented in the 
diagram accomplishes three functions: it prevenís a mixing of behaviors; it 
ensures that all necessary behavior systems get their turns; it also guarantees 
that the organism performs, at any given time, the activity most needed at 
the moment." (Translated from the Germán) 

Occasionally the physiological mechanism just described can have a 
dysteleonomic effect when it causes a very quick alternation of two con- 
flicting motor patterns. Even in ourselves we can, under certain harass- 
ing circumstances, observe alternating intention movements when we 
feel "torn between" two conflicting motivations. In some fish, intention 
movements alternating between attack and flight become very marked 
when two individuáis of the same species are positioned opposite one 
another and threatening. Each of them does the opposite of what the 
other is doing at the moment, and the interaction between their inten¬ 
tion movements can give rise to a self-amplifying oscillation. 

Like other non-teleonomic byproducts of neural organization, this 
oscillation can, by way of ritualization, attain the function of a signal. As 
Marler and Hamilton pointed out, the oscillation between fleeing and 
attacking has been ritualized in many small songbirds to form autono- 
mous motor patterns functioning as a signal of alarm or as one releasing 
the mobbing of a predator (1966). Fire-mouth cichlids, Cichlasoma meeki, 
often perform a ritualized ceremony when positioned opposite one 
another and going backward and forward across the border marking 
their territories. This ritual is different in certain characteristics from the 
unritualized actions from which it is derived. For one, it is beautifully 
rhythmic in a way that always makes the initiated observer suspect a 

4. Displacement Activities 


degree of ritualization. This is confirmed if one of the fish suddenly loses 
interest and turns away while the other continúes to oscillate backward 
and forward, thus demonstrating the autonomy of the new motor pat- 
tern. After the female pigmy cichlid, Nannacara anómala , has decided on 
a nest site, she attracts the male first by attacking him and then by switch- 
ing to the behavior pattern used to guide him to the nest. At higher 
intensities, an alternation of these two movements can escálate into a 
rather irregular oscillation. In the stickleback, the beautifully rhythmic 
"zigzag dance" of the male has originated in the same manner from the 
alternating intention movements of attacking the female and guiding 
her to the nest. This origin, however, is not immediately apparent 
because the ritualization has progressed much further than it has in 

4. Displacement Activities 

When two conflicting motor patterns are activated simultaneously, it can 
happen that the organism performs a third pattern which may belong to 
an altogether different system. This curious effect was first described in 
avocets, Recurvirostra avisetta, by G. F. Makkink, the Dutch ornithologist 
who called the movements thus elicited "sparking-over activities" (Über- 
sprungbezvegungen) (1936). The accepted English translation is "displace¬ 
ment activities." Although "displacement" means something entirely 
different in the psychoanalytic literature, this term, displacement activi¬ 
ties, will be used here. Some time later and independently of one 
another, Tinbergen (1951) and Kortlandt (1955) both studied this phe- 
nomenon. Tinbergen says: 

An examination of the conditions under which displacement activities usu- 
ally occur led to the conclusión that, in all known cases, there is a surplus of 
motivation, the discharge of which through the normal paths is in some way 
prevented. The most usual situations are (1) conflicts of two strongly acti¬ 
vated antagonistic drives; (2) strong motivation of a drive, usually the sexual 
drive, together with lack of external stimulation required for the release of 
the consummatory acts belonging to this drive. 

Elsewhere Tinbergen says that displacement activities also occur when 
the normal outlet for a certain motivation is "blocked." As a stickler for 
terminology, I must emphasize that the function of "blocking" here 
under discussion must not be confused with that of the blocks marked in 
Tinbergen's diagram of the hierarchical organization of instincts (Two/ 
IV/3). Using the terminology of this diagram, one would have to say that 
a certain motivation must become "de-blocked" and subsequently "re- 
blocked" by some obstacle situated a little farther "downstream" along 


VII. Múltiple Motivation in Behavior 

Figure 28. Explanation in text. (Czihak, Langer, Ziegler in Biologie.) 

the path of stimulus conduction. The excitation of an antagonistic activ- 
ity is one special case of such an obstacle; the lack of a stimulus situation 
necessary for the release of a final consummatory act is another. 

To explain displacement activities physiologically, A. Sevenster-Bol, P. 
Sevenster, and J. van Iersel have developed the following hypothesis. 
Each of two motivations activating incompatible behavior patterns exerts 
an equally inhibitive influence on a third one. Being elicited simultane- 
ously and with equal intensity, they neutralize each other, not only in 
regard to the activation of the motor patterns which they elicit normally, 
but also in regard to their inhibiting effect on the third action pattern. 
Thus the latter is disinhibited whenever these two motivations are active 
with equal forcé. This "disinhibition theory" of the displacement activity 
has been represented by B. Hassenstein (1976) in the diagram shown as 
Figure 28. 

An argument for the correctness of this theory is the fact that the effect 
of the stimulation, which normally releases the "autochthonous" activ¬ 
ity, can be added to that of disinhibition. Sevenster-Bol investigated this 
additive effect in the bilí shaking of Sandwich terns, Sterna sandvicensis. 
The activities of incubating as well as those of escape inhibit bilí shaking 
in these birds. Autochthonously, this action pattern is released by small 
objects, such as drops of water, adhering to the bird's bilí. In the conflict 
between alarm and incubation, bilí shaking appears regularly as a dis¬ 
placement activity; autochthonously it is performed with equal regular- 
ity when a fine drizzling rain is falling. Sevenster-Bol first counted the 
displacement activities performed in a standardized situation of conflict, 
then those observed without any conflict during a rain, and, lastly, those 
in the same situation of conflict plus the rain. The result was a very neat 
addition of both effects. 

4. Displacement Activities 


Another argument in favor of the disinhibition hypothesis is the fact 
that a vast majority of motor patterns appearing as displacement activities 
are common, "everyday" activities, in other words, are motor patterns 
that, because of their abundant endogenous impulse production, are con- 
stantly available. The so-called comfort activities of birds and mammals, 
such as scratching, preening, and shaking, furnish the most common 
examples of displacement activities; when embarrassed, even humans 
tend to scratch behind an ear—and in other places. Locomotion, too, is 
often disinhibited in conflict situations; some people tend to walk rest- 
lessly to and fro while delivering a speech, and so on. In all of these cases 
the disinhibition theory fits convincingly. 

Yet some typical and indubitable displacement activities have proper- 
ties which the disinhibition hypothesis fails to explain. The intensity of 
the motor patterns appearing as displacement activity should, according 
to the theory, be dependent only on the equilibrium between the two 
motivations canceling each other, but we know a great number of exam¬ 
ples of displacement in which the intensity of the displacement activity 
is clearly correlated to the intensity with which the two conflicting 
motivations are excited. In any case, the displacement activity should 
never reach levels of intensity surpassing those reached in autochtho- 
nous motivation. If one approaches the nest of black-capped warblers, 
Sylvia atricapilla, the birds mob the apparent predator with loud alarm 
calis and the alternating movements of turning towards and away from 
the enemy, as described in Two/VII/3. This ritualized and, in this situ- 
ation, autochthonous behavior is constantly interrupted and suppressed 
by bouts of violent preening and wing shaking. These activities, nor- 
mally occurring after bathing, reach an intensity never observed in their 
autochthonous performance. The intensity borders on frenzy and gives 
the impression of something pathological. The same is true for other dis¬ 
placement activities, and some of these will be mentioned in a subse- 
quent section. 

Another phenomenon left unexplained by the disinhibition theory is 
the high specificity of certain displacement activities that, although they 
are common, everyday activities, appear exclusively in the conflict 
between a particular pair of motivations and between no others. If the 
conflict between the incubating and the escape motivations gives rise to 
displacement bilí shaking in the Sandwich tern, we must ask why other 
comfort activities, equally common and equally suppressed by either 
incubation or escape, do not also appear in the conflict situation. I, 
myself, do not know a single case in which the same conflict causes more 
than one action pattern as a displacement. As the readiness for each of 
these activities is demonstrably fluctuating, a different one should appear 
at different moments, and it should be the one possessing the greatest 

Displacement activities are not only often characteristic of a certain 
conflict, but also of a certain species. In the same conflict between escape 


VII. Múltiple Motivation in Behavior 

Blocking through 
secondary inhibition 

'Sparking over" to the track 
of another motivation 


Displacement Hypothesis 

Figure 29. Functional diagram of the "displacement" hypothesis relative to dis¬ 
placement behavior. (Czihak, Langer, Ziegler in Biologie.) 

and nest defense, each of the ganders of three species belonging to the 
same genus perform a different kind of displacement activity. The grey- 
lag. Anser anser , performs a wing shaking, the intensity of which sur- 
passes by far that which the same movement ever reaches when it is 
autochthonously motivated; the pinkfoot. Anser brachyrhynchus, goes 
through the preening movements that distribute fat from the oil gland, 
located among the upper tail-cover feathers, over the flank feathers; the 
greater snow goose. Anser coerulescens atlanticus, goes through intensive 
bathing movements while remaining on dry ground. The correlation 
shown by the intensity of displacement with that of the conflict and also 
the specific dependence of a certain displacement activity on a certain 
conflict both argüe in favor of the original hypothesis expressed by Mak- 
kink (1936). The hypothesis suggested by the term "sparking over" con- 
tains some truth; it is represented in Hassenstein's diagram (Figure 29). 

Whether displacement activities have or do not have any teleonomic 
valué has repeatedly been discussed. Tinbergen and others have sug¬ 
gested that displacement activites may serve as a safety valve discharging 
superfluous motivation. Some displacement activities have been called 
by Tinbergen "an outlet of the fighting instinct." This interpretation is 
probably correct for those cases in which a sudden cessation of external 
stimulation leaves the animal with an excess of specific excitation which 
has to be got rid of in some manner. The displacement activities observed 
under these circumstances have been termed "after-discharge activities." 

I do not believe in the necessity for finding, at any cost, a teleonomy 
in displacement activities; it is altogether possible that they represent an 
a-teleonomic byproduct of neural organization. One argument for this 
supposition is the fact that displacement activities, through a process of 
ritualization, so very often attain new functions as signáis. Obviously it 
is advantageous to the species if the conflict within one individual can 
be communicated to the others. It is actually difficult to find examples of 
displacement activities that are not at least slightly changed by rituali- 

4. Displacement Activities 


zation; in other words, they have undergone changes under the selection 
pressure of their new function as "releasers" or as signáis. 

It has been suggested that in Germán, too, the term Übersprungbewe- 
gung —literally "sparking-over activity"—should be abandoned. It is cor- 
rect in principie to postúlate that the term chosen for a certain phenom- 
enon should not suggest an explanatory hypothesis. Still, as we have 
seen (page 252), Makkink's primary hypothesis seems to contain some 
truth, at least for certain cases. Ñames are inessential and, rather than 
create confusión through new terms, I propose, with admitted inconsis- 
tency, to retain the term Übersprung in Germán, and the equally unsat- 
isfactory term "displacement" when writing in English. 

Part Three 

Adaptive Modification of 


Chapter I 


1. Modification and Adaptive Modification 

External influences always effect the way in which an organism devel- 
ops. The genetic program, the genotype, stakes out the limits within 
which changes can be effected by environmental influences. Because the 
external circumstances that happen to influence each organism's devel- 
opment are never ever quite identical for two individuáis, the form in 
which an organism actually appears is never exactly the same as that of 
others with the same genotype. This individual form is called the phe- 
notype (from phainomai, "I appear" in Greek). The changes wrought by 
external influences, that is to say, the deviations of the phenotype from 
the genotype, are called modifications. The realization of any genetic pro¬ 
gram that has evolved during phylogeny is dependent on innumerable 
external conditions influencing the organism during its individual 
development, during ontogeny. A modification is any more or less last- 
ing change brought about in an organism by the circumstances of its 
environment during the course of its individual life. Modifications are 
omnipresent. There is hardly any small change in environmental con¬ 
ditions that does not cause a slight modification which may last only as 
long as a few minutes or as long as a lifetime. Two individuáis growing 
up in slightly diíferent environments will always show small differences 
in their phenotypes. 

A significant and widespread error is the assumption that any modi¬ 
fication brought about by a certain environmental influence must nec- 
essarily constitute an adaptation to that influence. Modification as such 
is not directed at adaptation any more than mutation and other genetic 
changes are. Modifications take place at random, although they are gen- 


I. Modification 

erally more determined and less stochastic than mutations are. Mutations 
brought about by the mutagenic poison, colchicine, are not adaptations 
increasing the organism's resistance to this poison, ñor are bones 
deformed by rachitis an adaptation to a deficiency of vitamin D or of 

If we find that a certain change in the environment regularly causes a 
modification which does constitute an adaptation to this change, we may 
assume, with an overwhelming probability, that this adaptedness is 
founded, as any other is, on phylogenetically acquired information. In 
other words, an adaptative modification is always the realization of a pro- 
gram which has evolved phylogenetically and which has been stored 
genetically. The modifiability of this program may be regarded as a spe- 
cies-characteristic adaptation to changes in the environment that are to 
be expected, but the direction of which is not predictable. If a mammaTs 
fur regularly grows thicker through exposure to a coid climate, or if a 
plant grows longer when illumination is decreased, or if a man acquires 
a greater number of red blood corpuscles when living high up in moun- 
tains where the oxygen contení of the air is diminished, these teleonomic 
changes are caused by external circumstances, but they are nonetheless 
the realization of a genetic program which has evolved as an adaptation 
to, and as a provisión for, the possibility of just these particular circum¬ 
stances. Verbalized, the information given to the plant in our example 
would be: When there is insufficient illumination the stem must be 
lengthened until the leaves receive enough light for photosynthesis. 

This kind of genetic instruction enabling the organism to gain mastery 
over various changes in an unstable environment has been termed, by 
Ernst Mayr, an open program (1942). The physiological mechanism under- 
lying this program enables the organism not only to acquire information 
not contained in the genes, but to retain and store it for a considerable 
time span, often for the duration of the individual^ life. The open pro¬ 
gram provides a finite number of possibilities; the stored information 
enables the organism to choose, from among the options contained in 
the program, the one that is teleonomically most promising under given 
circumstances. The choice of the teleonomically preferable option is a 
process of adaptation; the existence of the open program is a State of 

2. Analogous Processes in Embryogenesis 

The adapting function performed in the realization of an open program 
consists of receiving and exploiting information coming from external 
sources and correctly interpreting this information when opting for the 
teleonomically correct one among the finite number of possible adapta- 

3. Learning as an Adaptive Modification 


tions. Experimental embryology has thrown considerable light on these 
important processes. The ectoderm of a newt embryo can develop into 
simple external skin, into the neural tube, and consequently into the cen¬ 
tral nervous system, and it can, lastly, form the lens of the eye. In every 
one of its cells the ectoderm contains information necessary for building 
every one of these organs. Left to itself, that is, if a piece were cut from 
the prospective ventral side of the embryo and isolated, the ectoderm 
would form nothing but skin. If, however, it is placed in cióse proximity 
to a piece of the dorsal side of the primitive intestinal tract that later 
forms the notochord, the ectoderm will develop into a neural tube. When 
approached from the inside of the embryo's body by the growing eye 
cup, the ectoderm will be influenced to build a lens. Experimental trans- 
plantations have shown that influences emanating from these neighbor- 
ing organs " induce " the ectoderm to form the required structure. If, for 
instance, a small bit of prospective notochord is implanted under the 
embryo's ventral skin, the latter will develop a corresponding piece of 
neural tube. 

What happens in all adaptive modifications is, in principie, akin to the 
processes of embryogeny just mentioned. It does not matter whether the 
inducing influence is emanating from external factors or from adjacent 
organs within the embryo. What is essential is that the organism, or the 
organ, knows "from within," how to cope with a number of different 
eventualities and is informed, "from without," which of the eventualities 
it is that has actually arisen. 

3. Learning as an Adaptive Modification 

A vast majority of the processes which we are wont to describe as "learn¬ 
ing" effect a modification of behavior that is clearly adaptive. The 
unquestioned teleonomy of all learning processes forces us to assume a 
phylogenetically evolved program, just as it does in the examples of bod- 
ily modification mentioned in Three/I/1. Even the few malfunctions of 
learning point in the same direction. In this respect, all learning is akin 
to the embryogenetic processes discussed above: learning selects from 
many possibilities contained in an open program the one that seems to 
fit current circumstances best. Environmental influences furnish the 
information about which possibility this is. 

Most learning processes, but by no means all, differ from embryoge¬ 
netic induction by being reversible. Most of what has been learned can be 
forgotten or "unlearned." Karl Bühler, my teacher in psychology, 
actually included this characteristic in his definition of learning. There 
are, however, at least two kinds of important learning processes which 
cause permanent changes in the machinery of animal and human behav¬ 
ior. These are a) the so-called imprinting processes, and b) what psy- 


I. Modification 

choanalysts have termed traumatization. It must be emphasized that 
these processes are not caused by reinforcement and are not self-main- 
taining through continual relearning, as will be explained in Three/III/ 
5, 6. 

However one wants to define learning, the definition must inelude its 
teleonomic function. Surprisingly, even the most brilliant investigators of 
learning processes have apparently failed to notice that an explanation is 
needed for the indubitable fact that learning practically always results in 
an improvement of the teleonomic function of behavior. Not only psy- 
chologists, but the older ethologists, too, seem to have regarded this as 
a matter of course! The few exceptional cases in which the function of 
learning mechanisms miscarries reveal the existence of a highly teleon¬ 
omic, genetically programmed learning system—in those cases, for 
instance, in which the relief of tensión is afforded not by avoiding the 
noxious environmental situation but by means of narcotic effeets, or in 
cases in which a supernormal object is more reinforcing than the normal 
one. Both effeets can give rise to vices. Otherwise the only examples of 
non-teleonomic learning concern the "insatiable" curiosity of man, and 
even that may serve to sustain and sharpen the human ability to learn. 

This rather surprising neglect of an all-important question became 
obvious only when the heated discussions between ethologists and 
behaviorist psychologists (described in the Introductory History) were 
taking place. When D. S. Lehrman in his critique of ethological theory 
(1953) defended Z. K. Kuo's hypothesis that a mammal embryo could 
learn in útero and that the chick embryo could learn, while still in the 
egg, to peck by having its head moved up and down passively by its own 
heartbeat, I answered that in order to avoid the assumption of innate 
movements Kuo and his followers were obviously assuming the exis¬ 
tence of an "innate schoolmarm" who was teaching the animal what to 
do. Admittedly, this was intended as a reduction of the argument to 
absurdity, and it did take years for me to realize not only that an innate, 
in other words, a phylogenetically programmed learning mechanism 
must indeed exist, but that the major problems concerning all learning 
processes were condensed in the question of how these inbuilt teachers 
achieved the task of teaching the organism only teleonomic behavior and 
of discouraging it from any dysteleonomic actions. 

Strict behaviorists who negate the existence of any phylogenetically 
evolved programs of behavior and who regard the notorious nature-nur- 
ture problem as obsolete, regularly assert that the concepts of what is 
"innate" and what is "learned" are nonvalid, for the simple reason that 
the one can only be defined through an exclusión of the other (Hebb 
1953). The assertion in itself is correct, but the reason given is ridicu- 
lously wrong. Recognition of the fact that the innate teaching program 
must take part in any learning process does indeed preelude any dis- 

3. Learning as an Adaptive Modification 


junctive conceptualization of what is innate and what is learned, but does 
so for reasons which are the exact opposite of those for which Hebb and 
other behaviorists repudiated the "dichotomy" of "innate" and 
"learned." Their belief that learning processes must be involved in every 
kind of behavior is entirely erroneous; but conversely, there does not 
exist a single case of teleonomic learning which does not proceed along 
the lines prescribed by a program containing phylogenetically acquired 
and genetically coded information. 

The amount of genetic information is the greater the more complex 
the learning function is. It is hardly possible to quantify, even roughly, 
the amount of information the "built-in teacher" of a more complicated 
learning process must possess. However, I daré to assert that, could one 
count it in "bits," the genetic information needed to establish the neu- 
rosensory organization of an open program would prove to exceed by far 
the information necessary for the mechanisms of a purely innate pattern 
of behavior. The realization of a closed program can be likened to the 
building of a house using prefabricated parts whose special forms allow, 
unequivocally, but one way of putting them together. I have actually 
seen such houses—sitting on absolutely level lava terraces near Hilo on 
the large island of Hawaii. But I can hardly imagine any other place 
where a house could be erected with an equal parsimony of information. 
It is easy, however, to visualize the enormous amount of additional infor¬ 
mation that would be needed if an otherwise similar house were to be 
erected on uneven and irregularly sloping ground. An open program, 
with its faculty to take in and exploit external information, does not 
require less, but incomparably more programmed information. 

The principie on which phylogenetic adaptation and adaptive modi¬ 
fication cooperate to achieve teleonomic behavior is, I think, fairly easy 
to understand. This understanding, however, is essential to future eth- 
ological research. The most important question always is: Whence comes 
the information that makes the behavior teleonomic? In what elements 
and structures is it contained, and how does it act in guiding all learning 
to the goal of reinforcing teleonomic and of extinguishing dysteleo- 
nomic behavior? Whoever finds his life's work in this kind of problem 
will, with amazement and complete lack of comprehension, have to take 
into consideration the still widespread opinión that the conceptual dis- 
tinctiveness of phylogenetically programmed and learned mechanisms 
of behavior is "not only infertile, but actually false," as Robert Hinde has 
put it (1959). Equally amazing are the following statements made by 
Edward O. Wilson (1978) in his book, On Human Nature : "... even in the 
relatively simple categories of behavior we inherit a capacity for certain 
traits, and a bias to learn one or another of those available. Scientists as 
diverse in their philosophies as Konrad Lorenz, Robert A. Hinde, and B. 
F. Skinner have often stressed that no sharp boundary exists between the 


I. Modification 

inherited and the acquired" (page 60). Such statements are as surprising 
to the ethologist as they would be to a geneticist or to a student of pop- 
ulation dynamics, if someone were to tell them that the concepts of geno- 
type and phenotype are "not valid."* 

^Similar distortions of my findings and opinions can be found elsewhere in the oth- 
erwise admirable writings of E. O. Wilson, and in the publications of others. It is an easy 
and cheap tactic to ascribe to a scientist witless opinions that he has never held, and then 
to make him look stupid by disproving them. 

Chapter II 

Learning Without Association 

1. Facilitation and Sensitization 

The process of association, that is, forming a connection through indi¬ 
vidual learning between two or more experiences, was discovered and 
investigated by Wilhelm Wundt (1922). Since then this kind of process 
has become the focus of interest of almost all research concerned with 
learning. The conditioned response, which occupied I. P. Pavlov as well 
as some important American investigators at the turn of the century, is 
nothing other than the objective, physiological aspect of what Wundt 
called association. With some justification, the conditioning of responses 
was and still is regarded as the essense of all learning. 

Thus it is all the more necessary to State that the wide definiton of 
learning given in Three /I/3 ineludes a number of processes which def- 
initely achieve an adaptive modificaron of behavior, although associa¬ 
tion plays no part in them at all. On the other hand, the nonassociative 
learning processes to be discussed in this chapter do, indeed, cooperate 
often enough with association, and in this way form mixtures of, and 
seeming transitions between the two. 

One form of learning which can be considered the most primitive of 
all is the simple improvement of function by functioning, comparable to 
the "breaking-in" of a car or any other machine. The process of facilita¬ 
tion (Bahnung in Germán, meaning literally "breaking a trail") can be 
observed in several nervous functions, particularly in that of stimulus 
conduction. Facilitation is probably caused by changes in synapses and 
this would furnish a plausible explanation for the fact that the perfor¬ 
mance of some behavior patterns becomes smoother and faster after a 


II. Learning Without Association 

few repetitions. The process of facilitation must not be confused with 
that of maturation which, in growing organisms, can produce very sim¬ 
ilar effects. 

As M. J. Wells has demonstrated (1962), the prey-catching response of 
a newly-hatched squid (Sepia) is performed with perfect ease and flaw- 
less coordination when it is released for the first time, but much more 
slowly than after a few repetitions. The improvement of the squid's prey- 
catching response is very probably due not merely to facilitation but to 
maturation as well. According to the classical behaviorists' own defini- 
tion, the concept of operant conditioning (conditioning of the type R) is 
only applicable in those cases for which, in a situation new to an organ- 
ism, various modes of behavior are "tried out" and the successful one is 
retained. This is definitely not the case for the newly-hatched squid; in 
fact, the typical and unchangeable motor pattern of "shooting out" their 
tentacles is recognizable through intention movements several seconds 
before the action is actually initiated. In Hess's chicks (see below), it is 
equally unlikely that conditioning plays a role because its effects, if any, 
would tend to disturb their aiming mechanism rather than improve it. 
The diminishing of the scattering of their pecks certainly has nothing to 
do with information gathering by operant conditioning. When "train- 
ing" newly-hatched chicks (1956), E. Hess found an effect similiar to that 
demonstrated by Wells. Hess fitted his chicks with spectacles having 
prisms; these simulated a sideways shift of all visual objects. The chicks 
never learned to compénsate for this deviation. Their pecking consis- 
tently missed the goal on the same side but, none the less, with exercise 
the scattering diminished considerably. 

Processes similar to those of facilitation also take place in the sensory 
sector of the central nervous system. They are grouped together under 
the concept of sensitization. At the release of a certain behavior pattern, 
the threshold of its key stimuli is lowered. In other words, the "atten- 
tion" of the animal is awakened by the first response. This anthropo- 
morphic term already implies that, unlike the motor facilitation already 
discussed, this type of sensory facilitation is a very short-lived change in 
the machinery of behavior. 

Obviously, sensitization can perform a teleonomic function only when 
the first impingement of key stimuli permits, with sufficient probability, 
the prediction that other stimuli will be arriving subsequently. An earth- 
worm (Lumbricus) that, most likely due to the sensitization of the giant 
fiber escape response, has just avoided being eaten by a blackbird (Turdus 
merula) that has taken a peck at it, is indeed well-advised to respond with 
a considerably lowered threshold to similar stimuli, because it is almost 
certain that the bird will still be nearby for the next few seconds. 

In responses other than those of escape, as in those of catching and 
eating prey, sensitization for obvious reasons makes sense only when the 

2. Habituation or Stimulus Adaptation 


presence of one prey animal implies the probability of some others being 
about. As Wells (1962) has shown in impressive examples, marine ani¬ 
máis of different kinds, preying on species occurring in swarms or 
schools, are thrown into a State of extreme excitation by catching one 
comparatively small prey. The "feeding frenzy" thus induced makes 
sharks snap at the most inappropriate objects, and the same response in 
tuna (Thynnus) is exploited by commercial fishermen: after having 
snapped up a few pieces of bait, these fish will snap blindly at unbaited 

However, the "feeding frenzy" of these fish is a function on a slightly 
higher level. The catching of a single prey lowers the threshold of snap- 
ping in general. But still the process is much more complicated. It does 
not depend only on the eating of one prey animal; it also depends on 
what is often called "social induction": the stimuli emanating from vora- 
ciously feeding conspecifics increase appetite, and this feedback is 
mainly responsible for the frenzied escalation of a response that is inde- 
pendent of appetite. "Feeding frenzy" can be induced in fish that are too 
ill to eat normally, for example, and also in healthy fish through the pre¬ 
sentaron of uneatable objects. The stimulus configuraron that consists 
of many objects of equal size and similar shape massed in one location is 
what releases the typical darting about and even the aiming at one of the 
uneatable objects which, then, is rejected only at the last moment. 

2. Habituation or Stimulus Adaptation 

A learning process found in the lowest as well as in the highest metazoa 
is habituation to a certain stimulus situation. Habituation can be equated 
to a desensitization to a quite specific stimulus situation. The situation is 
thus deprived of the releasing influence it previously exerted. 

The fresh water coelenterate, Hydra, responds to a number of different 
stimuli by contracting its body as well as its tentacles. It responds to 
movements of the surrounding water, to touch, to a shaking of the sub- 
strate, and so on. Nevertheless, a hydra can live in a brook sufficiently 
turbulent that its body and arms wave constantly to and fro. The motion 
of the water has obviously lost its releasing effect and it can make the 
hydra sway passively without causing contractions. 

Habituation has been termed sensory adaptation by some European 
physiologists and has been compared by them to sensory fatigue. This is 
misleading because habituation is, quite on the contrary, a physiological 
mechanism whose teleonomy consists in preventing fatigue. Yet, I like 
to speculate that the mechanisms of habituation have arisen phylogenet- 
ically by a process of "narrowing in" fatigue to very small sectors in the 
afferent part of a behavior system. M. Schleidt, trying to create an unpre- 


II. Learning Without Association 

judiced term, spoke of a "throttling down" of afference. Fatigue of the 
entire response is avoided by confining the waning of function to spe- 
cific afferent pathways. What is important about the hydra's habituating 
to the stimuli emanating from flowing water is the fact that the thresholds 
of all other key stimuli eliciting the same response of contraction remain 
unchanged. While the animal is fluttering and waving about in the cur- 
rent, its reactions to the touch of an enemy or of prey remain as finely 
triggered as they ever were when the hydra was hanging motionless in 
still water. 

As stated above, some physiologists have accepted the term "sensory 
adaptation" ( Sinnesadaptation in Germán) as a synonym for habituation. 
This term should be abandoned for two reasons. For one, habituation 
occasionally produces unadaptive dysteleonomic effects; for another, the 
term suggests that the process takes place in sensory organs only, as does 
the light-and-darkness adaptation of the retina. The process here dis- 
cussed not only produces more long-lasting consequences than does sen¬ 
sory adaptation, in the strict sense; it is, moreover, by no means confined 
to the sensory organ itself, but demonstrably takes place in much more 
central parts of the nervous system. Margret Schleidt investigated the 
"gobbling" response of the turkey cock (1954). The IRM releasing this 
motor pattern is not very selective; quite a number of key stimuli proved 
to be effective. This made it possible to investigate habituation to one 
particular kind of stimulus. A short sound, produced by an electronic 
generator and repeated at constant intervals, at first elicited one gobbling 
response each time it was offered. If stimulation was continued patiently 
for a longer period, the gobbling gradually became rarer and rarer until, 
at last, it was uttered with no greater frequency than it would have been 
in vacuo, for example, by a turkey sitting alone in a soundproof chamber. 

When the pitch of the sound was changed, the response reappeared at 
once, and it was possible to demónstrate that the desensitization con¬ 
cerned only a very narrow range of sound frequencies extending just a 
very little above and below that of the original stimulus; on each side of 
the peak, the curve of adaptation fell off very steeply. The thresholds to 
sounds whose frequencies were more markedly different from that of the 
primary stimulus, remained entirely unchanged. The same was true for 
all sounds produced by other instruments. 

These phenomena could still be explained by an "adaptation" or 
fatigue in the sensory organ itself. However, this assumption was dis- 
proved by an extremely simple experiment. After the bird had become 
completely habituated to a certain sound, Margret Schleidt subjected her 
turkey to sounds of the same pitch and duration, at the same intervals 
and produced by the same generator, but of a much lower amplitude, 
that is, less loud. Surprisingly, this change completely abolished the 
desensitizing effects of the preceding habituation; the turkey responded 

2. Habituation or Stimulus Adaptation 


to the soft tone exactly as it would have done to any other entirely new 
stimulus situation. This experiment completely precluded the possibility 
of the process being localized in the sensory organ itself. Also, it cannot 
be related to any sort of fatigue, otherwise the diminished intensity of 
stimulation could never have resulted in a new resurgence of response. 

Chapter III 

Learning Through Association 
Without Feedback Reporting 

1. Association 

Wilhelm Wundt (1922) defined associations as " . . . connections between 
contents of consciousness which . . . possess the general character of non- 
intentional processes, taking place in a passive state of conscious atten- 
tion" ( " . . . Verbindungen zwischen Bewu/ftseins-Inhalten die . . . den 
allgemeinen Charakter unzvillkürlicher , bei passivem Zustand der Auf- 
merksamkeit eintretender Bewu/3tseinsvorgánge besitzen"). Associations 
are produced when two events happen at once or several times in the 
same sequence and within short intervals of time. From this Wundt for- 
mulated the laws of "succedaneity" and of "contiguity." Both are valid 
for a majority, if not for all of the learning processes that will be 
described here. 

The welding together or linking up of two subsequent subjective and, 
accordingly, physiological events has the important consequence that 
the organism, after having experienced the first, has come to "expect" 
the second, that is, has come to prepare for it. I. P. Pavlov's dog began to 
salivate when it heard the sound of the little bell it had come to associate 
with being fed. The experimenter in a laboratory intending to study 
association naturally applies the stimulus he wants to condition shortly 
in advance of the unconditioned stimulus, that is, he lets the bell ring 
before he lets the dog have its food, or before he uses another stimulus 
that "unconditionally" elicits salivation. The regularity of this sequence, 
artificially produced by the researcher, should not induce us to forget 
that, under natural conditions, an equally reliable sequence of two things 
happening one after the other occurs only when there is a causal connection 
between them. In the mountains of Armenia I often watched herds of 

2. Habituation Linked with Association 


half-feral goats gallop wildly towards a sheltering cave whenever they 
heard the thunder of an approaching storm—a highly teleonomic behav- 
ior initiated in preparation for the subsequent downpour. It was dyste- 
leonomic, however, whenever the goats did the same thing after hearing 
the noise produced by prisoners of war dynamiting rocks. 

There can be no doubt about the faculty for association having evolved 
as an adaptation to the laws of conservation of energy. Energy can take 
diíferent forms, although always in a strictly lawful sequence. Through 
its function of preparing the organism for something to be expected in 
the near future, the faculty for forming associations is analogous to the 
category of causality in human thought processes. The similarity, even 
the identity of function has misled great philosophers such as Hume into 
thinking that they are one and the same. Hume's famous problem arises 
from this confusión. Logically, the fact that the sun has risen on every 
day since the beginning of time is in itself not a logical argument in favor 
of the assumption that it will rise tomorrow. It is a built-in mechanism 
of an altogether diíferent kind that forces us to assume a causal connec- 
tion between two or more events after they have happened a few times 
in the same regular sequence; the category of causality, independent of 
and often contrary to logic, compels us to make this assumption. A lack 
of logic does not prevent the causal assumption from being teleonomi- 
cally correct in an overwhelming majority of cases. Like other mecha- 
nisms exploiting instant information, it hardly ever errs! 

2. Habituation Linked with Association 

As direct observations of higher animáis in their natural habitats show, 
there are processes of habituation linked with, and dependent on, very 
complex functions of perception. These functions quite certainly are per- 
formed at higher levels within the central nervous system. As every 
aviculturist as well as every trainer of animáis knows, certain responses 
are bound to a general environmental situation that consists of innu¬ 
merable stimulus data, none of which may be missing if the response is 
to remain undisturbed. These stimulus data combine to form a "complex 
quality," as Karl Bühler called it. The complex quality of stimulus data 
provides a multifaceted image of the momentary environmental situa¬ 
tion and also a context for the múltiple interrelationships. A very slight 
change in the general environmental situation is often sufficient to 
"break down" an animal's usual and expected response. For instance, a 
bird that appears to be absolutely fearless of man, taking food from his 
hand without the slightest hesitation, will be thrown into a panic if the 
man approaches from the "other" side. A dog that has first learned to lie 
down on an uphill pathway and responds to the command reliably in 


III. Learning Through Association 

this situation, must also learn that this command is just as obligatory 
when going downhill on the same path. Often the habituation to a com- 
plex quality breaks down at small changes not at all obvious to the 
human observer. If a bird is thrown completely out of phase because the 
sand covering the bottom of its cage is of a slightly different color, one 
is in doubt whether to consider the creature as very clever for noticing 
the change, or as very stupid for being upset or confused by it. 

From the physiological point of view, habituation linked with associ¬ 
ation is a puzzling process: a certain configuraron of stimuli loses its pre- 
vious releasing function after having arrived a number of times together 
with a multitude of other data, which in their turn are selectively per- 
ceived and integrated as a single complex quality by the mechanism of 
gestalt perception discussed in One/II/3. This waning of the original 
response is sustained only as long as all of the concomitant stimuli inte¬ 
grated into the complex quality arrive together. If any one of them is 
missing, the complex perception breaks down and the primarily eífective 
"unconditioned" stimulus situation regains in full its former releasing 
function. A very complex performance of perception has, through asso¬ 
ciation, become the prerequisite for the functioning of the otherwise 
simple process of habituation. We do not know how this is achieved 
physiologically, but we are familiar with many examples. What is gen- 
erally meant by saying that an animal has become tame is, essentially, 
that it has become habituated to previously flight-eliciting stimulus sit- 
uations associated with perceiving the nearness of human beings. 

Two examples may serve to illustrate the high specificity of this kind 
of habituation. When we moved our wild geese and other Anatidae to 
the newly-established institute on the Ess-See, we feared that our birds 
would become the prey of marauding foxes—beause they were com¬ 
pletely unafraid of our externally somewhat fox-like chow chow hybrids, 
to which they were habituated. As it turned out, their habituation was 
associated exclusively with those individual dogs. Even a strange chow 
chow belonging to a visitor was mobbed, and the reaction to foxes was 
even stronger. Also, the habituation to our own dogs broke down when 
we arranged for the birds and dogs to be together outside the institute 
area on the other side of the lake. 

A second example of habituation associated with a complex situation 
is furnished by the aggressive behavior of shama thrushes (Copsychus 
malabaricus). These birds drive the young of an earlier brood out of their 
territory when those of the next brood are fledging. Once I wanted to 
keep a young male beyond that time and, in order to protect it against 
the expected attacks by its parents, I shut it up in a cage. As if by a "creep- 
ing in" of stimulation, the older birds habituated to the inevitable pres- 
ence of this son and paid no attention to the cage and the bird in it— 
until I made the error of moving that cage to another córner of the same 
room. From that moment the habituation disappeared completely and 

2. Habituation Linked with Association 


both parents attempted to attack the captive bird; they would have let 
the younger brood die of starvation if I had not removed the disturbance. 

In some instances, situation-associated habituation can have dysteleon- 
omic effects. There are a number of responses whose teleonomic function 
is quite obvious but which still tend to wane after a few repetitions so 
that, strictly speaking, their full teleonomic effect is accomplished only 
at their first elicitation. 

In the chaffinch (Fringilla coelebs), the warning-and-mobbing response 
directed at owls loses more than half of its intensity after a few repeti¬ 
tions and, as Robert Hinde has shown, fails to be renewed after a rest 
period of weeks and even months. According to Hinde, this waning can- 
not be explained by the absence of reinforcing stimulation; even by 
allowing the experimental bird to be chased by a real little owl and by 
having a few tail feathers torn out in the process, the chaffinch's warn- 
ing-and-mobbing response could not be restored to its original intensity. 

The response of hand-reared greylag goslings to an imitation of the 
parent's warning cali showed the same uncontrollable waning. It is 
inconceivable that responses possessing so clear a teleonomic function, 
such as the two just mentioned, have evolved only to function but once 
during an individual^ life. For this reason we must conclude that there 
is something wrong with our experimentation. Possibly part of the wan¬ 
ing is due to the experimenter's tendency to keep the conditions "con- 
trolled and constant"; yet comparable constancy can occur in nature and 
I find it impossible to believe that a chaffinch begins to consider a little 
owl harmless after having seen it twice or three times sitting on the same 
branch. Perhaps we are ruining the responses by our impatience, releas¬ 
ing them diligently in much too quick a succession—something that 
would never occur in the wild. However, Hinde certainly did not com- 
mit this particular error, and the rapid waning of highly teleonomic 
responses remains a riddle. 

The cooperation between habituation and the associated processes of 
perception can convey to the organism a very particular form of infor- 
mation. Unlike the ordinary function of habituation, the process here 
under discussion does not tell the organism that a certain situation is 
harmless, but on the contrary, calis the animal's attention to a particular 
danger. Habituation allied to association here develops a function which 
is otherwise performed by sensitization. As W. Schleidt has shown 
(1961), the wild turkey's response to flying predators is released by an 
extremely simple combination of key stimuli. Everything appearing in 
dark silhouette against a lighter background above the bird and moving 
with a certain angular speed, that is, a speed correlated to the size of the 
object, releases intense escape reactions in wild turkeys of all age groups. 
A fly creeping along the white ceiling of the laboratory releases the same 
response as a balloon floating across the sky, or as a helicopter, or as a 
hawk (Buteo buteo). As dummy experiments revealed, the outline of the 


III. Learning Through Association 

object was absolutely irrelevant, but habituation to any particular form 
of the dummy set in very quickly. Any object that had been shown a few 
times lost most of its effectiveness, and any other object would surpass 
its releasing valué. The strongest releasing objects were always those 
which the birds had not seen for the longest time. The free-ranging tur- 
keys at our institute in Seewiesen gave the strongest escape responses to 
a dirigible which, rented by an advertising firm, flew its round ¿ver the 
vicinity about twice each year. Among all of the objects tested, buzzards 
exerted the weakest releasing influence because they were seen daily 
flying above our area. This was so in spite of the fact that, of all the 
objects investigated, they were the ones most similar to the bald eagle 
(Haliaeetus leucocephalus), the only bird of prey that is a real danger to 
adult turkeys. The information given to the turkeys through the process 
just described could be verbalized thus: Among the objects slowly glid- 
ing across the sky, the most dangerous is the one which is seen least often. 
This would unequivocally characterize the bald eagle. 

3. "Becoming Accustomed" or Habit Formation 

There is a particular form of learning that is, in a manner of speaking, 
the antithesis of habituation: while habituation causes an originally 
effective combination of key stimuli to lose its releasing function when 
associated with the perception of a more complex configuraron, the pro¬ 
cess to be described here has the contrary effect. Key stimuli, originally 
sufficient in themselves for the release of a certain response, become 
associated with a multitude of other stimulus configurations and, from 
then on, retain their function only when presented within the context of 
that complex configuration. Because the behavior thus becomes depen- 
dent on primarily non-releasing situations, the process could well be 
called "habit formation"—if this term were not burdened with the con- 
notations of drug addiction. In common parlance, it means to accustom 
a person to a thing, "becoming accustomed to something" that, during 
the process, becomes more and more indispensable. In Germán, the verb 
angewóhnen has exactly this meaning. 

The process is very common and can also be described by saying that 
an IRM is rendered more selective by an association between innate 
reactions to key stimuli, on one side, and conditioned responses to com- 
plicated gestalt perceptions on the other. It must be emphasized, how- 
ever, that by getting accustomed to a conditioned stimulus combination, 
the latter is not transformed into a substitute for the original key stimuli— 
as it is in forming conditioned responses. The key stimuli to which an 
IRM responds remain indispensable even when the associated gestalt per¬ 
ception becomes equally indispensable. Henceforth, both can elicit the 
response only when acting together. W. Schleidt has coined the term 

3. "Becoming Accustomed" or Habit Formation 


EAAM (E from erzverben = acquire, AAM for Angeborener Auslósemechan- 
ismus = innate and acquired releasing mechanism) for this combination 

An increase in the selectivity of an IRM through the formation of habit 
is a very common process and has been demonstrated in humans, too. As 
René Spitz has shown, the smiling response of the human baby can ini- 
tially be released by an extremely simple dummy—a toy balloon with 
eyes and a grinning mouth painted on it. A nodding movement is a 
strong additional key stimulus. The effect of the stimuli obeys the law of 
heterogeneous summation (Two/I/14); the configuration of the eyes 
seems to be most important. The effect of the nodding movements is 
enhanced by a color difference between the face and the top of the head. 
For a time Spitz, who was bald, wondered why his nodding was so much 
less effective than that of his dark-haired assistant, until he realized the 
cause and put on a dark cap. 

When smiling first appears, these innate key stimuli are as effective 
depicted on a crude dummy as they are on the face and head of a live 
person, but after the baby has smiled only at humans and not at dummies 
for a few weeks, it begins to fear the dummy and to give the smiling 
response only to human beings. When the baby has reached the age of 
five or six months, the smiling response has become even more selective 
and can be released only by a person familiar to the baby, usually its 
mother. The baby now reacts to strangers by showing fear and by trying 
to turn its face away from them. 

Even in humans the process of "becoming accustomed" just described 
does not render the original key stimuli either ineffective or dispensable. 
Even when the mother is fully recognized by the baby as an individual 
person, it will still smile at her with the greatest intensity when she 
bends over it, smiles at it, and nods her head. 

In his paper on the history of the concept of IRM (1962), Wolfgang 
Schleidt has thoroughly analyzed the way in which IRMs are rendered 
more selective by learning. Otto Storch termed this process "receptor 
learning" and called attention to the fact that it appears in phylogeny at 
much earlier stages than does any motor learning (1949). Schleidt says: 

The filtering effect of the IRM is often enhanced during ontogeny by the 
learning of additional characteristics or by habituation to stimulus configu- 
rations which have been encountered repeatedly. I propose to sepárate con- 
ceptually from the IRM (in a strict sense) those IRM "modified by experi- 
ence" and to use the acronym IRME. Releasing mechanisms which either 
have lost the originally existing structure of their IRM, or which have devel- 
oped without an underlying IRM, can be separated from the previously 
mentioned types as "Acquired Releasing Mechanisms" (ARM). If there is no 
experimental basis for such a classification, or if it is irrelevant whether the 
linkage between stimulus and response has been established by phyloge- 


III. Learning Through Association 

netic or ontogenetic adaptation to characteristics of the environment, the 
term "Releasing Mechanism" (RM) should be used without an additional 

I must repeat that this process of increasing the selectivity of IRMs 
through learning is extremely common, not to say omnipresent. In fact, 
IRMs which are not adaptively modified by experience, thereby having 
achieved a higher degree of selectivity, are not easy to find in higher 

That phase in a baby's life during which it becomes accustomed to con- 
tact with individuáis and during which it learns to refuse a stranger as 
a substitute for someone known, represents one of the most important 
periods in human ontogeny. The very first formation of a bond with an 
individual is the basis and the prerequisite for the later development of 
the faculty to form bonds of human love and friendship. A number of 
organs and a number of behavior systems are susceptible to an extremely 
rapid process of atrophy if they are not used almost at once upon their 
ontogenetic emergence. In humans, the faculty for establishing personal 
bonds is one of them. During this critical period, as René Spitz has dem- 
onstrated, hospitalized children begin to form a personal attachment to 
the nurses taking care of them. When the routine turnover of hospital 
personnel deprives them of these mother figures, attempts to develop a 
bond with the substitute nurses is already noticeably weaker, after still 
another turnover weaker still, until, at last, the wretched children give 
up, turn their faces to the wall and refuse to initiate any further attempts 
at establishing contact. René Spitz has produced a truly heartrending 
film of this process. In extreme cases, the poor babies simply die, but 
even if less damage is done and a child survives and later learns to 
negotiate with other human beings, its emotional structure is perma- 
nently damaged. All of its emotions are eerily blurred and the faculty to 
love, to become attached to personal friends, is severely diminished. As 
personal love is the most important antidote against aggressivity, the lat- 
ter is often dangerously uninhibited in these unfortunate children. Also, 
curiously enough, their exploratory behavior is very often atrophied 
together with the reduced capacity for bond formation. A characteristi- 
cally empty, disinterested facial expression is a symptom of these patients 
and, in some respects, they resemble those suffering from infantile 
schizophrenia, with which the phenomena under discussion are often 
confused. René Spitz subsumed them under the concept of hospitalization. 

Habit formation, in the sense of becoming accustomed to a stimulus 
configuration to the point of making it indispensable , plays a very impor¬ 
tant part in the formation of social bonds in many different animáis, as 
well as in man. For those animáis in which the process is limited to a 
strictly definable period during ontogeny, as is the first bond formation 
in humans, it bears a certain resemblance to the process of imprinting that 

3. "Becoming Accustomed" or Habit Formation 


will be discussed in Section 6 of this chapter. One example of the simi- 
larity is this: a newly hatched, inexperienced greylag gosling first reacts 
by "greeting," and a little later by following, any moving object which 
gives, in response to its "lost piping," a series of rhythmically repeated 
sounds within a certain range of pitch. After having "greeted" in this 
manner a human being two or three times, the gosling refuses to react 
in the same way to any other object, its real mother included. The irre- 
versibility of this "object fixation"—as Freud would cali it—is character- 
istic of imprinting. 

One of the great mysteries of imprinting is to be found in the fact that 
it fixates behavior onto the species and not onto the individuality of the 
object. Once the gosling's following response has been imprinted on 
humans, the gosling cannot be made to follow a goose, but the human 
individual fostering it can be exchanged for another without any dimin- 
ishing of the following response. Also, a gosling hatched by its real 
mother occasionally changes over to another goose family and remains 
with that family. 

Imprinting of the following response is succeeded by a process of habit 
or custom formation. During a timespan of roughly twenty hours, the 
gosling does not reliably recognize its parents as individuáis. Interest- 
ingly enough, the parents' voices are recognized slightly sooner than are 
their physical forms. The process of getting to know the parents is 
demonstrably independent of reward or punishment; if a gosling hap- 
pens to wander away from its family soon after leaving the nest, it will 
try to join any strange family with goslings of approximately the same 
age. If the latter are more than two days oíd, their parents will take 
exception to the little stranger and bite it, in a mild and inhibited man¬ 
ner, but still strongly enough to make it utter distress calis and to run 
away. If one reunites the wayfarer with its own family, it shows that it 
has definitely not profited by its disagreeable experience; on the con- 
trary, goslings which have once joined a strange family tend to repeat 
this error. It is as if even a short following in the wake of the "wrong" 
parents tends to blur the image of the true ones. 

In the process of learning to recognize its parents, the gosling's IRM, 
which releases all filial responses, becomes associated with one of the 
most complex functions of gestalt perception. In ourselves, the analogous 
faculty of visually recognizing our fellow humans as individuáis is 
largely dependent upon our perception of the configurations of eyes, 
eyebrows, and nose. It is surprising how effectively the covering of this 
area impedes recognition; the small, conventional carnival mask is suf- 
ficient to do so quite effectively. Curiously enough, it is the same portion 
of the head which is essential for personal recognition among geese; a 
sleeping goose with bilí and forehead tucked under its wing becomes 
completely unrecognizable to its fellows and is occasionally bitten by 
mistake. One of the funniest sequences is that in which a gander, having 


III. Learning Through Association 

thus bitten his beloved mate, recoils with astonishment and switches to 
abject greeting patterns. Fledged, full-grown goslings, temporarily sep- 
arated from their parents, search for them most assiduously and, while 
doing so, respond optimistically to any goose that is not positively iden- 
tifiable because its head is hidden under its wing or under water—as if 
it could be one of the lost parents, they rush up to it greeting intensely, 
and start back disappointed when the head of a stranger appears. Hand- 
reared goslings are perfectly capable of transferring their mechanism of 
facial recognition to the human foster parents in spite of the enormous 
differences of body proportions. After a time, and somewhat longer than 
that taken for parent-reared goslings to learn individual recognition, the 
filial responses of hand-reared goslings, such as greeting, following, and 
snuggling up, can be released exclusively by the foster mother, no matter 
how she is dressed. It is the image of the face alone that is relevant, as 
the following response is in no way diminished if the body of the 
human, unlike that of a goose, becomes invisible when swimming. 

4. The Conditioned Reflex Proper or Conditioning with 
Stimulus Selection 

The term "conditioned response" has been used by I. P. Pavlov and many 
others rather comprehensively to inelude some learning processes which 
are very much more complicated than the one to be described here. As 
Hassenstein has pointed out, many American behaviorists inelude 
within this all-embracing concept learning mechanisms that are very dif- 
ferent from one another. 

Following Plassenstein, I shall limit the term "conditioned response" 
to a designation of the result of a simple association connecting an orig- 
inally ineffective stimulus with a reaction that can be regarded as a 
"reflex" insofar as its function is not subject to changes of internal readi- 
ness (as was discussed in One/I/2). Blinking an eye, a "reflex" which 
serves to protect the cornea, can be caused to occur by optical stimulation, 
for instance, by quickly moving an object towards the eye, or by tactile 
stimuli, such as blowing a fine jet of air against the cornea of an open 
eye. It is quite impossible to suppress the blinking reaction by willing 
the eye to stay open. 

The time which elapses between stimulus and response, the so-called 
reaction time, varies between .25 and .40 sec. Acoustic stimuli, if not of 
overpowering loudness, do not release blinking. If the soft tone of a 
buzzer is made audible a number of times just before the optical or tactile 
stimulus is applied, the nervous system responds to this repeated expe- 
rience by establishing a new reflex connection: the tone of the buzzer 
alone is henceforward able to elicit the reaction, independent of subse- 
quent optical or tactile stimulation. The central nervous system has, in a 

4. The Conditioned Reflex Proper or Conditioning with Stimulus Selection 277 

manner of speaking, taken cognizance of the fact that the acoustic stim¬ 
ulus precedes the releasing stimulus in time with a certain regularity, so 
that the one may be relied upon to announce the other. On the basis of 
this "assumption," a new "wiring" of the reflex is achieved; this shortens 
the reaction time by putting the reflex on notice and preparing it for 
what is to come. Hassenstein says: 

On the basis of experience, a new connection is established. The following 
cogent indirect conclusions are to be drawn from experimental observations. 
Graphically, they are depicted in [Figure 30]. Between sensory elements and 
executing organs, the learning process has formed a new and relatively sta- 
ble signal-conveying connection. The temporal sequence between the signáis 
sent by the conditioned and by the unconditioned stimulus must have 
caused the formation of a new and comparatively permanent connection between 
the two pathways. There must be, within the central nervous system, some 
loci which, through their sensitivity to temporal sequences in the arrival of 
signáis on two originally independent pathways, are capable of building a 
lasting new signal-conducting path between them. If the central nervous 
system does not possess this kind of inbuilt mechanism, a conditioned reflex 
cannot be developed. For instance, it is impossible to connect a tendón reflex 
with a conditioned stimulus. 

. . . Except on the phenomena just described, the conclusión is founded on 
no other presupposition than that no physical distance effects are at work 
within the organism. 

According to the notion just presented, the information concerning prec- 
edent experience is stored in the brain by a process in which two signáis, as 
well as their aftereffects, act simultaneously and at the same location and that 
the result of this is the formation of a new signal-conducting pathway. As 
far as I can see, no other hypotheses are conceivable. The question how new 
connections are developed within the central nervous system remains as yet 

The passive evaluation of the sequence in which the stimuli arrive per- 
forms its teleonomic function quite particularly under conditions and in 
environmental situations which the organism cannot change by its own 
activity. Skinner (1938) and others describe this learning process, in 
which a new hitherto ineffective stimulus is selected, as conditioning of the 
type S, in contrast to those other learning processes by which not a stim¬ 
ulus, but a successful behavior is selected. This latter is called operant con¬ 
ditioning, or conditioning of the type R. Conventionally, the concept of 
the conditioned reflex also ineludes the so-called elassie conditioning 
represented by the dog's conditioned salivation reflex first investigated 
by I. P. Pavlov. 

As shall be discussed in the section on conditioned appetence (Three/ 
IV / 3), Pavlov abstracted a law that is perfectly valid in very many cases 
from observations of behavior in which it does not prevail; in the history 
of Science, quite a number of analogous, productive errors have occurred. 


III. Learning Through Association 

® es)-— 


signáis in 

l/neural pathways 

New signal-conducting 
connection formed ¡n 
consequence of coincidence 
of arriving signáis 

Figure 30. Idealized and simplified flow diagram of the processes indispensable 
for stimulus conduction and storing of data in the formation of a conditioned 

reflex. (CS = conditioned stimulus, US = unconditioned stimulus; US - R = 

original unconditioned reflex, CS- R = conditioned reflex.) Phase 1 represents 

the state before learning: no connection exists between the stimulus-to-be-con- 
ditioned and the reaction R. Phase 2 illustrates the simultaneous arrival of two 
physiological signáis at the locus sensitive to coincidences in the central nervous 
system. Phase 3 characterizes the functional state of the new system. The condi¬ 
tioned, as well as the unconditioned stimulus is, by itself, capable of releasing the 
conditioned reaction. (Hassenstein, B.: Information and Control in the Living 

5. Avoidance Responses Acquired Through Trauma 

There are certain acquired avoidance responses which are akin to the 
conditioned reflex as conceived by Hassenstein. One such response is 
acquired when an originally indifferent, non-releasing stimulus situa- 
tion is transformed into the conditioned stimulus by impinging on a pas- 
sive organism at the same time that an unconditioned stimulus releases 
strong escape reactions. What is characteristic of this process, which 
indubitably represents a conditioning of the S type, is that a single expe- 
rience is sufficient to create the association, and, secondly, that the con¬ 
ditioning is irreversible—probably the effect of the shattering, unforget- 
table, terror-eliciting nature of the experience. The whole complex 
quality of the concomitant circumstances under which it was made 

6. Imprinting 


becomes the "conditioned stimulus." In this respect the process under 
discussion resembles the learning processes described in Sections 1 and 
2 of this chapter. 

Conditioned avoidance responses can be observed even in very simple 
organisms; phylogenetically they have probably evolved from the pro¬ 
cesses of sensitization discussed in Three/II/1. In some flatworms (Plan- 
aria), stimulation by a light does not primarily elicit any escape response, 
but it can be shown to have a mild sensitizing effect on a subsequent 
reaction to the unconditioned stimulus of shaking the substratum. By 
repeatedly applying these two stimuli in succession, the light stimulus 
can be made into a true conditioned stimulus releasing avoidance 
responses. By some stretching of the imagination and of the conception, 
this can be regarded as a transition between sensitization and association. 
In higher animáis and in humans, a superlatively affecting, terror-inspir- 
ing stimulation often becomes irreversibly associated with the accom- 
panying situation, thus working irremediable damage. Psychiatrists cali 
this effect a trauma. 

The conditioned stimulus situation which henceforward elicits uncon- 
trollable terror, can be of varying complexity. One of my dogs once got 
caught in a revolving door. Because of this single experience she not only 
meticulously avoided revolving doors wherever they happened to be, 
but she also fought shy of the whole vicinity of the building in which 
the mishap had occurred. When forced to pass through the Street on 
which the building was located, she crossed to the opposite sidewalk and 
flashed by with her ears laid back and her tail between her legs. Trainers 
of dogs and horses, like psychoanalysts, know all too well how inerad- 
icable conditioned avoidance responses of this type can prove to be. 

6. Imprinting 

Another learning process, which we cali imprinting, is similar to the true 
conditioned response because it is based on association alone, but, on the 
other hand, it also resembles traumatic conditioning because it is irrever¬ 
sible. Here the association between a certain behavior pattern and a cer- 
tain stimulus situation also takes place without reinforcement, that is, 
without the feedback of some rewarding effect of an action. In other 
words, it represents another example of type S, or conditioning with 
stimulus selection. One very important criterion of imprinting is its lim- 
itation to a circumscribed ontogenetic phase during which the organism, in a 
manner of speaking, "waits" for some very definite unconditioned key 
stimuli to arrive, only to associate them instantly with a certain part of 
the accompanying stimulus situation. A phylogenetic program deter¬ 
mines precisely when the young organism is to learn what. After hatch- 
ing and becoming able to look around, a greylag gosling utters its lost 


III. Learning Through Association 

piping, to which under normal conditions its mother answers with a 
rhythmic cackle. To this the gosling responds by "greeting." The 
mutually releasing sequence of piping-cackle-greeting is predictable 
with a high degree of probability and the moment at which it will first 
take place is equally predictable. The program timing the period of sen- 
sibility of that irreversible learning process for this particular moment is 
obviously adaptive. The inbuilt learning mechanism conveys to the 
gosling information which, if verbalized, would say: "When you first 
feel lonely, utter your lost piping, then look out for somebody who 
moves and says 'gang, gang, gang' and never, never forget who that is, 
because it is your mother." An enormous amount of data is integrated 
into the releasing mechanism by this process of irreversible instant 

The teleonomic function of imprinting is to form an irreversible con- 
ditioning of a response to its biologically adequate object. Sigmund 
Freud, who discovered the process independently of ethologists (its first 
discoverer, according to Eckhard Hess, was the English monk, Saint 
Cuthbert, 635-687), spoke of "object fixation" (1905, 1916-1917). In most 
cases of imprinting, a fellow member of the species is the object of the 
response and the process, thus, is one of the mechanisms guarding 
against hybridization. In other cases, imprinting can irreversibly fixate 
reactions on other objects; for instance, the prey-catching responses of 
predatory birds to a certain species of prey animáis. 

The temporal sequence of the sensitive periods during which different 
behavior mechanisms become fixated on their objects shows no correla- 
tion to the sequence in which these Systems later mature. In jackdaws, 
the sensitive period for sexual imprinting comes quite a while before that 
for imprinting the following response. In this bird, sexual imprinting 
takes place while the nestling sits in the nest still half-naked, showing 
no responses to fellow members of its species except that of gaping at its 
parents when they arrive to feed it. The imprinting of the responses for 
following the parents has its sensitive period immediately before fledg- 
ing. The effect of sexual imprinting does not become apparent until two 
years later, while the imprinting for the following response is observable 
after only a few days. 

The irreversible association effected by imprinting always concerns 
one behavioral system. It should never be said that a bird is imprinted to 
humans; it is always only one response which is thus fixated. As early as 
1935 I stressed that different responses of the same individual can be 
imprinted to very different objects. The jackdaw just mentioned was sex- 
ually imprinted to humans; its reactions of flocking together were fixated 
on hooded crows (Corvus corone); its behavior of parental care responded 
normally to young jackdaws. One of the reasons greylag geese are such 
a glorious object for sociological studies is that their filial and social 
responses can be imprinted to human beings without affecting in any 
way their choice of object in sexual behavior. 

6. Imprinting 


One of the unexplained properties of imprinting is the association of 
the unconditioned reaction with a stimulus situation that, though infi- 
nitely more complex than the configuration of key stimuli releasing the 
original IRM, characterizes the species of the adequate object in a generic 
manner. A bird such as a mallard (Anas platyrhynchos) or a zebra finch 
(Taeniopygia castanotis) reared in the company of a bird of another, related 
species, will become imprinted for the purposes of sexual responses to 
that species but not confined to the individual bird in whose company 
it was reared. Mallard drakes reared with male sheldrakes by Fritz Schutz 
(1965) and, after a few weeks, liberated on the Ess-See, flocked with other 
mallards and showed none of the usual sexual responses to their own 
species. Instead, in the following spring, they started to court male shel¬ 
drakes. In no case did they court the individuáis with which they had 
been reared and which had effected their imprinting. This may have 
been due to chance because, even if they did not recognize a former com- 
panion, this foster sibling was but one among a dozen male sheldrakes 
on the lake. Still, it is possible that the incest-preventing mechanisms, 
highly developed in most waterfowl, may have played a role. A jackdaw 
I hand-reared and thereby imprinted sexually to humans, never 
responded sexually to me, but it fell violently in love with a petite, dark- 
haired girl. The surprising thing was that the jackdaw "regarded" her as 
belonging to the same species as myself. 

Some biologists have tried to interpret imprinting as a conditioning by 
reward, searching assiduously for some stimuli acting as reinforcement. 
Their rather artificial and complicated hypotheses are not convincing; 
instead, they show how strong is the ideological bias of the belief that 
learning by reinforcement is the only learning process extant. In the 
zebra finch (Taeniopygia castanotis), Klaus Immelmann has shown incon- 
trovertibly that imprinting and learning by reinforcement are two dif- 
ferent processes (1965). He demonstrated their independence by setting 
them against one another. Young males, fostered by Bengalese finches 
(Lonchura bengalensis) , were later given the choice between females of 
their own species and those of Lonchura. They regularly chose the 
imprinted Lonchura species for mating. When Immelmann arranged the 
rearing of another group of zebra finches by Bengalese, and then kept 
the males exclusively in the company of females of their own species for 
weeks and finally for months, the males ultimately mated with their own 
kind. After these males had successfully reared several broods with con- 
specific females, Immelmann put Bengalese females in the enclosures 
with the mated couples. None of the experimental birds left his zebra 
finch mate immediately, but after they had been exposed to the 
imprinted objects for some time, they gradually left their previous mates 
and changed over to Bengalese females. The imprinting process, 
although followed by no reinforcement whatsoever, vanquished the 
effects of the strongest kind of reinforcement imaginable. The same phe- 
nomenon was demonstrated in an even more dramatic manner by my 


III. Learning Through Association 

friend, B. Hellmann, working with budgerigars (Melopsittacus undulatus). 
He brought together a male and female, both of which were sexually 
human-imprinted, in a large room where they had no occasion to see 
humans again since they were fed and watered through a chute. It is well 
known that, when budgerigars are deprived of normal sexual objects, 
they easily turn to substitutes, henee the celluloid dolls or mirrors often 
found in their cages. So it was to be expected that these birds, deprived 
of the object on which they were sexually fixated, would finally find 
adequate substitute objects in each other. This they did; they successfully 
mated and reared their young. Later, just as they again had a brood of 
young, the birds were exposed to contacts with humans. Hellmann and 
I entered the garret room. The male and the female rushed towards us 
immediately and began to court us violently. Although we immediately 
withdrew, the pair began to fight and subsequently, through neglect, 
they allowed their babies to starve to death. The engram acquired 
through imprinting evidently cannot be effaced by any other learning 
process, not even by the most effective kind of reinforcement known to 

The effect of imprinting is ineradicable, but so is the phyletically 
evolved function of IRMs, and the two phenomena can give rise to a 
conflict, as Schutz has demonstrated (1965). When he was experimenting 
with mallards (Anas platyrhynchos ), he found that males could easily be 
imprinted sexually to other species, while in females such imprinting 
was apparently ineffective. Even when Schutz exposed the females to the 
strongest possible imprinting situation by allowing them to be reared by 
foster mothers together with foster siblings of another species, they 
invariably mated with mallard drakes after having been liberated on a 
lake. Schutz was quick to suspect that this was not due to the absence of 
any imprinting effect, but to the overwhelming strength of key stimuli 
emanating from the mallard drake's nuptial colors and courtship move- 
ments, to which the female's IRMs were keyed to respond. By means of 
a very elegant experiment, he proved this suspicion to be correct. It is 
known from the investigations undertaken by H. Mau (1973) that mal¬ 
lard females, even when given huge doses of male hormones or even 
after having been spayed, cannot be impelled to perform male courtship 
movements. The immediately visible effect of testosterone injections is a 
great increase of female courtship activities, very like that of estrogen 
injections, but with one important difference: the estrogen-injected birds 
directed their courtship activities toward males, while those of the tes- 
tosterone-injected females were invariably addressed to females. In other 
words, it is only the choice of the sexual object which is directly influ- 
enced by hormones in the female mallard. Schutz imprinted seven mal¬ 
lard females to muscovy ducks (Cairina moschata). Subsequently, he kept 
these ducks in a large aviary in the company of an equal number of mal¬ 
lard drakes and muscovy drakes. All the muscovy-imprinted mallard 

6. Imprinting 


Figure 31. Duration and course of the imprinting phase in ducklings. Every duck- 
ling was exposed for one hour to an artificial dummy. Afterwards it was deter- 
mined in an experiment involving choice whether imprinting on this object had 
occurred or not. (Hess in Eibl-Eibesfeldt, I.: Ethology: The Biology of Behavior.) 

females mated normally with conspecific drakes. After this had hap- 
pened, Schutz implanted the mallard females with testosterone crystals, 
whereupon all of them left their mallard mates and began violently to 
court the muscovies, to whom they had paid no attention at all before. 
As long as these mallards reacted as females, they were susceptible to the 
releasers of male mallards, but the moment their hormones commanded 
them to choose a female as a sexual object, they responded according to 
the imprinting they had undergone, exactly as male mallards would. 

A majority of investigations of the imprinting process have been per- 
formed on the following response of young, autophagous, precocial 
birds. In the mallard, Eckhard Hess found that the sensitive period for 
the imprinting of this behavior extends from shortly before to shortly 
beyond the fifteenth hour after hatching, but showing a distinct peak at 
the fifteenth hour (Figure 31). 

Many so-called imprinting experiments have been performed on 
domestic chicks and have led to highly misleading results because, in 
this species, the following response is not fixated on its object by typical 
imprinting at all. The imprinting of the following reaction seems to be 
suitable for research merely because the eggs of domestic chickens and 
mallards are available at any time. But the most important criterion of 
imprinting, its irreversibility, can hardly be demonstrated in a response 
which begins to wane from the first day and which, in chicks and mal- 


III. Learning Through Association 

lards, disappears completely after a few weeks, in geese after a few 
months. Furthermore, as the response in these animáis begins to opérate 
so soon after hatching, it is difficult to prove that no reinforcement is 
conditioning the following activity. 

It should always be remembered that imprinting was discovered in the 
sexual responses of birds: again and again Heinroth, myself, and, ages 
ago, Spalding and von Pernau, were disappointed by the impossibility 
of breeding hand-reared birds; they persisted in courting the aviculturist 
and persistently refused to pay any attention to members of their own 
species. In sexual imprinting, the unimportance of reinforcement is 
made obvious by the fact that the fixation on a certain object is often 
accomplished more than a year before the response can be released. 

There is a superficial resemblance between the imprinting of the fol¬ 
lowing response and the conditioning of appetitive behavior (which will 
be discussed in the next chapter); the sequence of events participating in 
both processes is the same. The young bird hears the cali note of its par- 
ents, or sees the parents' movements, or both, and reacts with the instinc- 
tive system of following. While doing so, it learns to recognize the indi- 
viduality of each parent. To this extent, the prerequisites for forming a 
conditioned appetence are certainly given. There are, however, reasons 
to doubt that the satisfaction of a need for contact with the parents really 
acts as a reinforcement. There have been many debates about imprinting, 
and also some experiments conducted to explore it. Many of the latter 
have been performed on the mother-following response of domestic 
chicks, as noted above. Although easily obtained, this response is not at 
all ideal for a study of the imprinting process. I repeat that the most 
important characteristic of imprinting, its irreversibility, so blatantly 
obvious in sexual imprinting, can hardly be demonstrated in a mother- 
following response of chickens when, normally, this response lasts only 
a few weeks and is actually on the wane from the first day onward. E. 
Hess, in his book (1975), gives an excellent résumé of the work done on 

7. Conditioned Inhibition 

Association alone—that is, association without feedback effects, which 
are to be discussed in the next chapter—can achieve still another learn¬ 
ing process of a particular kind: association is able to create the inhibition 
of an activity even before it has occurred. As every trainer of dogs 
knows—or should know—it is quite impossible to dissuade a dog from 
poaching by punishing it after it has come back. Doing that creates a 
conditioned aversión not to poaching but to coming back. If, on the other 
hand, the trainer learns to anticipate the dog's running off—one 
becomes only too sensitive to the dog's tense attitude and to the partic- 

7. Conditioned Inhibition 


ular slinking trot indicative of hunting intentions—and to punish it at 
once, for instance, by throwing a bunch of keys at the offender, it is per- 
fectly possible to produce an association between the punishment and 
the intention to initiate a certain behavior, in other words, a conditioned 
inhibition to running away. Even a very slight punishment occurring 
simultaneously with the offense has a much stronger and more lasting 
effect than any extinguishing stimuli, however strong, that impinges 
after the act. In other words, conditioned inhibition is much easier to 
establish than the conditioned avoidance to be discussed in IV/4. 

The intentional production of conditioned inhibitions plays an impor- 
tant role in the training of domestic and of circus animáis. Most "teach- 
ing" by extinguishing or punishing stimuli acts on the same principie, 
particularly the so-called breaking of horses, which is customary in some 
parts of the New World and which appears so atrociously cruel to any- 
body familiar with the Spanish Riding School and its methods. 

Hassenstein has called attention to a possible danger incurred by the 
forming of conditioned inhibitions. Under their influence a behavior 
pattern will occur, if at all, much more rarely than is normal, and this 
can cause an equally abnormal increase in its motivation, particularly if 
the inhibited pattern represents the only outlet of a certain ASP. Some 
accidents with circus animáis are definitely the outcome of this "dam- 
ming up" of action-specific potential. E. Zimen has observed in wolves 
(Canis lupus) (1971) and H. van Lawick in African hunting dogs (Lycaon 
pictus) that an individual that has been suppressed for a long period by 
a dominant pack member, and that has remained passive for a very long 
time, can suddenly break out in a desperate attack against the suppressor. 
All mechanisms which normally check aggressivity, such as the reaction 
to appeasement gestures, remain completely ineffective in that kind of 
attack, which in the case observed by H. van Lawick, very nearly resulted 
in the death of the former tyrant. In humans, as Anthony Storr (1968) 
has reported, the killing of a husband or a wife is by far the most fre- 
quent type of murder, and is most often committed by men or women 
who, up to the moment of the crime, had been abjectly submissive to 
their wives or husbands and, according to Storr's interpretation, it is the 
very intensity of love, inhibiting even the mildest utterance of aggres¬ 
sivity, that leads to its catastrophic internal and unnoticeable accumula- 
tion, and results in a final and tragic eruption. 

Hassenstein has furnished a convincing cybernetic interpretation of 
conditioned inhibition (1965, 1966). He first delineates the variables of 
input and output: 

The input variables are represented first, as in the case of conditioned action 
[Three/IV/5], by the central nervous impulse for the activity later to be 
inhibited; second, as in the conditioned aversión [Three/IV/4], by the mes- 
sage reporting the disagreeable experience as a consequence of which the 


III. Learning Through Association 

conditioned inhibition has been developed. The output variable is repre- 
sented by the activity suppressed by the learning process just discussed. 

The formation of a conditioned inhibition is different from that of a 
conditioned reflex (in Hassenstein's sense, as described in Section 4 of 
this chapter) in one important aspect: the conditioned stimulus is not 
associated with a passive readiness to a reaction, not with a "reflex," but 
with a central nervous impulse, with a "command" in the parlance of 

8. Chapter Summary 

1. In all the learning processes discussed in this chapter, the adaptive 
modification of behavior is caused by the formation of a connection, 
by an association , between a stimulus situation which primarily has no 
releasing effect—the "conditioned stimulus"—and a second configu¬ 
raron of stimuli—the "unconditioned stimulus"—which acts as a 
"key stimulus" on a IRM and thus activates a certain behavior pattern. 
The learning process achieves the choice of a stimulus situation in which 
this behavior pattern performs a teleonomic function. 

This learning process corresponds to B. F. Skinner's concept of con- 
ditioning of the type S, which is characterized by stimulus selection 
(1936). According to Skinner, the following law prevails in this pro¬ 
cess: The approximately simultaneous presentation of two stimuli — 
one of which (the reinforcing stimulus) forms part of a reflex that, at 
a given moment, is developed with a given forcé—can result in a 
strengthening of another reflex, the increase consisting of the 
response to the reinforcing stimulus and to the originally releasing 
stimulus. As Foppa (1965) suggests, this is just another description of 
classical conditioning. Skinner also formulated a law of extinction. If 
a reflex which has been strengthened by a conditioning of the type S 
is elicited a number of times without being followed by the reinforc¬ 
ing stimulus, its strength will diminish. 

2. This latter, rather vague definition fits the one type of learning pro¬ 
cess discussed in Section 1 of this chapter: habituation is a process by 
which a stimulus configuration—often a very complicated one— 
becomes associated with the key stimulus of an IRM in such a manner 
as to "extinguish" the effect of the key stimulus. This complete 
suppression of originally effective key stimuli lasts only as long as the 
complex of the associated configuration endures unchanged. The elim- 
ination of any small, seemingly unimportant detail of that complex 
can quite suddenly re-establish the full eíficiency of the key stimuli. 

3. "Becoming accustomed" to a particular configuration of stimuli is a 
process that, in a manner of speaking, is reciprocal to the process of 

8. Chapter Summary 


habituation: the originally effective key stimuli are associated with a 
complex configuration in such a manner that they retain their eífec- 
tiveness only when presented together with that complex. 

4. In the formation of the conditioned response—in the strict sense 
defined by Hassenstein—and also in the formation of an avoidance 
response brought about by a "trauma," association has an exclusively 
unilateral effect: the releasing valué of the originally effective "uncon- 
ditioned" key stimulus is in no way changed while, by being associ¬ 
ated with it, an originally non-releasing combination of stimuli 
acquires an efficacy equal to that of the unconditioned stimulus. 

5. Like the process of "becoming accustomed" to a complex of stimuli 
configurations, imprinting results in an increase of the selectivity of 
an IRM; like the formation of an avoidance response after a trauma, 
imprinting is irreversible. The most important function of imprinting 
is to fixate certain activities on their biologically adequate object, in 
most cases on a fellow member of the species. 

6. Conditioned inhibitions are formed when a behavior pattern, at the 
very moment of its emergence, is countered by a punishing stimula- 
tion such as fright or pain. The process becomes teleonomic by sup- 
pressing a behavior under the circumstances under which it was fol- 
lowed by these undesirable consequences. 

All the paragraphs of this chapter have dealt with learning processes 
that induce an animal to perform—or not to perform—a particular action 
upon the arrival of a certain previously irrelevant stimulus, which is the 
principie of stimulus selection or of conditioning of the type S. In other 
words, the animal learns by having something happen to it. In the next 
chapter I shall discuss learning processes in which the animal gains 
information by doing something itself. It should be mentioned here, how- 
ever, that not all processes in which an animal gains information 
through its own activities represent conditioning of the type R, or oper- 
ant conditioning. I advócate that this term should be applied only to pro¬ 
cesses in which the organism learns to select, among the several behavior 
patterns in its repertory, the one fitting the immediate situation. In a vast 
majority of cases, the animal gains information by trying to apply one 
particular behavior pattern to different environmental situations and 
learns, by this activity, to choose the stimulus situation in which this par¬ 
ticular activity affords a máximum of reinforcing feedback. Although the 
organism is active throughout, this is not a process of behavior selection 
but a process of stimulus selection. While this type of process—teaching 
the animal the teleonomically correct situation in which to discharge a 
phyletically programmed activity—is extremely common, the type of 
learning that really conforms to the accepted definition of type R con¬ 
ditioning is comparatively rare in nature. 

This type R process is regarded by most learning theorists as the very 


III. Learning Through Association 

essence of all learning, by some as the only one worthy of investigation, 
and by a few as the only important element of behavior altogether. Even 
Foppa (1965), in his brilliant exposition on the learning theories of 
important American psychologists, underrates the importance of the 
learning processes discussed in this chapter. He says: "For the natural 
process of behavioral adaptation, conditioning of the type R is much 
more important." In contradiction to this, I must State that even rather 
highly developed organisms, such as greylag geese, rely on learning pro¬ 
cesses with stimulus substitutions for practically all the necessary adap- 
tive modifications of their behavior. Actually, I had some difficulty find- 
ing an example of operant conditioning in this species, and a rather 
doubtful one at that. When learning what to eat, the gosling does indeed 
apply different motor patterns while it is nibbling, in an exploratory 
way, on some new object, and it does indeed achieve a behavior selection 
of sorts, although there are only a few and closely related motor patterns 
to choose from. In corvid birds, in which exploratory behavior plays an 
incomparably more important role, this is quite different, as will be dis¬ 
cussed in Chapter VI. Over-assessing the importance of operant condi¬ 
tioning and underestimating all other types of learning probably stems 
from a generalization of properties inherent in the learning processes of 
the rat (Epimys norvegicus) and of man (Homo sapiens): both are extraor- 
dinarily exploratory creatures, and the importance of operant condition¬ 
ing correlates directly with exploratory behavior. 

Chapter IV 

Learning Effected by the 
Consequences of Behavior 

1. The New Feedback 

Without any known exception, animáis that have evolved a centralized 
nervous system are able to learn from the consequences produced by 
their own actions, success acting as a "reward" or "reinforcement," fail- 
ure acting as a "punishment" tending to "extinguish" the animal's readi- 
ness to repeat the action just performed. This learning process cannot be 
explained even hypothetically on the basis of an association between two 
pre-existent neural systems. An "open program" with a much more com- 
plicated structure must be postulated. As discussed in Two/I/1, the Sys¬ 
tem which Heinroth called the arteigene Triebhandlung, species-character- 
istic drive action, consists of several physiologically distinguishable 
components; a) appetitive behavior, b) achieving a stimulus situation to 
which an IRM responds and, finally, c) a consummatory act, in which a 
phylogenetically programmed action is performed, by which a teleon- 
omic function is achieved and the "drive" or motivation of the action is 
stilled. This three-link chain of processes is the base on which the ability 
to learn by success or failure has evolved phylogenetically; also, it still 
remains an indispensable part of the system achieving this kind of learn¬ 
ing. Wallace Craig had fully grasped this when, in his classic paper, 
"Appetites and Aversions as Constituents of an Instinct" (1914), he 
described the way in which the performance of the consummatory act 
induces the animal to seek a very special and complicated stimulus situ¬ 
ation in which the act aífords a máximum of satisfaction. The teleonomy 
of teaching the organism to do the right thing hinges on a feedback mech- 
anism. The stimulus configuraron giving satisfaction to the consumma¬ 
tory act must be characteristic enough of the situation in which this 


IV. Learning Effected by the Consequences of Behavior 

behavior achieves its teleonomic function in order to be able to afford a 
sufficiently reliable report on success or failure. This report must reach as 
far back as the mechanisms of precedent appetitive behavior, encourag- 
ing what has led to success and extinguishing that which has resulted in 
failure. Among behaviorist-learning theorists, it was only P. K. Anokhin 
(to the best of my knowledge) who postulated that, in the "conditioned 
reflex," a feedback cycle must be at work, a cycle in which the consum- 
mation and the adaptive success of the action is reported back and fed 
into the mechanisms of initial behavior (1961). As far as conditioning 
with behavior selection is concerned, he was perfectly right, though not 
with respect to all other learning processes conventionally but confus- 
ingly subsumed under that concept. 

Obviously, feedback reporting the success or failure of a behavior pat- 
tern is highly teleonomic and the fact that so many kinds of metazoa 
have, independently of each other, "invented" learning mechanisms of 
this type, just as they have invented eyes, fins, and other analogous 
organs, is, therefore, not so surprising. In fact, all phyla have done so, as 
far as they achieved the evolution of a centralized nervous system. 
Platyhelminthes, annelids, gastropods, echinoderms, cephalopods, crus- 
taceans, insects, spiders (Arachnomorpha), and vertebrates have done so. 

As was explained in the last part of the previous chapter, two kinds of 
conditioning are distinguished according to behaviorist nomenclature. 
In the first, the animal is not faced with a problem which it actively tries 
to solve; it learns by a purely passive experience—that is, a certain, in 
itself irrelevant stimulus can be taken as a sign that another relevant 
stimulus will presently arrive for which certain preparations can be 
made. This is type S conditioning. 

In sharp contrast to type S conditioning of stimulus selection, type R 
is defined as a process in which not a stimulus, but a behavior pattern is 
selected. The animal does not wait passively for a stimulus to arrive, but 
strives actively to master a certain stimulus situation which constitutes a 
problem. The animal learns, by acting, which of its several behavior pat- 
terns is adequate for the given situation. This is operant conditioning of 
type R. The distinction is, as Foppa has pointed out (1965), neurophys- 
iologically legitimate, as it certainly makes a basic difference whether a 
central nervous excitation is caused by the organism's moving in space, 
or by something else doing so. 

Although this clear and disjunctive conceptualization seems to brook 
no intermediates, we get into difficulties when we try to apply it to the 
learning processes we most often observe in wild animáis living in nat¬ 
ural surroundings. In all learning by success or failure the animal is doing 
something to get adaptive information; to this extent all of these pro¬ 
cesses conform to the definition of operant conditioning. However, it is 
extremely rare that they result in the selection of a behavior pattern. In 
a vast majority of cases it is a certain stimulus situation that is chosen. 

1. The New Feedback 


and to this extent the processes correspond to the definition of type S 
conditioning. The most common and quite certainly the biologically 
most important processes of learning by experience are those that form 
conditioned appetence (to be discussed in Section 3 of this chapter), and 
conditioned appetence for quiescence (Three/IV/4,5). Through both of 
these processes the animal learns from its own actions, but these actions 
remain the same throughout the process and it is a certain stimulus sit- 
uation that is selected by learning. 

In my opinión it is necessary to make a much sharper distinction than 
is generally done between the selection, by learning, of a behavior and 
the selection of a certain stimulus situation. Not all learning through the 
feedback of any activity undertaken is here considered operant condi¬ 
tioning. Operant conditioning is here equated with the selection, by 
"learning," of which behavior pattern, amongst many other ones, to 
choose in a given situation. It is not operant conditioning if, under the 
motivation of the appetite to perform one behavior, an animal learns by 
feedback which environmental situation is adequate. Ñor is it explor- 
atory behavior if the naive animal starts by performing this behavior pat¬ 
tern in an entirely "wrong" situation. A dog that "wants" to perform the 
behavior of burying a remnant of food (see One/II/5) first attempts to do 
so on the parquet flooring of the dining room, subsequently outdoors, 
and through a number of rewarding feedbacks it learns to choose the 
teleonomically correct place. It does not learn which behavior pattern to 
choose if it has an oíd bone left over; it learns to recognize the circum- 
stances under which to carry out phylogenetically pre-adapted motor 
patterns. In the same way a bird that "wants" to carry out the beautiful 
motor pattern of nest building does not learn how to perform the move- 
ments (which will be described later), but learns to recognize the situa¬ 
tion in which performing the nest-building movements gives the máxi¬ 
mum satisfaction, and so on and so forth. In all these cases it is the 
stimulus that is selected and not the behavior that is programmed to func- 
tion in a particular situation; ñor is the behavior of the dog or the bird 
true exploratory behavior, although, in both cases, the organism gains 
information by its activity. Exploratory behavior is founded on a very 
particular and highly complex program and is the subject of Chapter VI. 

Under laboratory conditions and, for instance, when constrained 
within a puzzle box, very many animáis try to escape by going through 
the whole repertory of behaviors at their disposal and learn to prefer the 
one that has led to success. In this case one can speak of true operant 
conditioning, that is, of behavior selection as opposed to the selection of 
a stimulus situation fitting one particular programmed behavior pattern. 
In nature, operant conditioning in this strict sense is much rarer than is 
generally supposed. 

Under normal conditions it is rare that an animal learns to choose, in 
a certain situation, a certain behavior pattern in preference to several oth- 


IV. Learning Effected by the Consequences of Behavior 

ers which are also at its disposal. Exploratory behavior is a highly dis- 
tinctive process that occurs only in the most highly organized animáis. 
It consists in the animal's responding to an unknown object by applying, 
tentatively, most or all of the programmed behavior patterns available to 
it. The motivation for this is absolutely autonomous and can, in subjec- 
tive phenomenology, be described as curiosity. In exploration the animal 
treats any new environmental situation "as if" it were biologically rele- 
vant. When exploring a new object, a raven performs motor patterns of 
flight, predation, eating, and so forth. A rat in a new environment 
explores all the possible pathways and hiding localities. Information 
gathered through exploratory behavior is not immediately obvious to 
observation; it becomes apparent only if and when the animal has 
recourse to an object fulfilling some present need; henee the accepted 
term of "latent learning." Exploratory learning is very important in the 
life of many higher animáis and quite particularly in that of humans. 

For a long time the research interests of behaviorists have been con- 
fined to operant conditioning. In humans and in some other highly 
explorative animáis such as the rat (Epimys norvegicus), this type of learn¬ 
ing does indeed predominate and is indeed worthy of the most thorough 
investigation. Any criticism directed at behaviorists cannot concern their 
brilliant analysis of operant conditioning, but only their neglect of the 
many other learning mechanisms in existence. 

In their study of operant conditioning, behaviorists have discovered a 
number of laws which are valid for all learning by success or failure. One 
phenomenon is found to be true for all conditioning by success or fail¬ 
ure: the association between the operant and the reinforcement does not 
endow the conditioned stimulus with an absolute and invincible releas¬ 
ing power, it only "facilitates" or "favors" the operant being performed, 
in other words, it makes its appearance more probable. This has been 
proved statistically by behaviorists and it is also understandable from the 
ethological point of view. As discussed in Two/II/7, the readiness to per- 
form a fixed motor pattern fluctuates within wide margins, as do the 
threshold valúes of key stimuli releasing it and also the appetitive behav¬ 
ior directed at its performance. The rewarding and reinforcing eífect 
exerted by the "satisfying" discharge of the action pattern is, of course, 
proportional to the present level of its ASP. This holds equally true when 
the action is represented by the "operant" in conditioning of type R, as 
when it is the appeted action pattern in the learning processes discussed 
in Chapter III. 

Everything learned under the influence of a reward or a punishment 
is gradually forgotten again if either of these two unconditioned eífeets 
fails to occur a certain number of times; the conditioning must then be 
inforced again, henee the "re-" in "reinforce." Neither in English ñor in 
Germán is there a word that describes this effect naturally. The first pub¬ 
licaron by I. P. Pavlov was written in Germán and the term he used was 

2. Mínimum Complication of the System 


"verstárken" (as D. Kaltenháuser very helpfully found out for me). In 
my opinión, the English expression most naturally describing the process 
under discussion would be "encouraging"—encouraging the animal to 
perform a rewarded action, or "discouraging" it if the "expected" result 
fails to be achieved. The conventionally accepted word "extinguishing" 
most unfortunately suggests a sudden disappearance of the learned 

2. Mínimum Complication of the System 

Unlike most behaviorists, ethologists persist in harping on the problem 
of how to explain the fact that, except in a very few illuminating dyste- 
leological cases, learning always ends up improving the teleonomic effect 
of the behavior it modifies. An apparent exception consists in the seem- 
ingly neutral and redundant learning done by humans that continúes 
through life and does not noticeably modify behavior. The process of 
this information gathering certainly comes under the heading of "latent 
learning." Humans as well as rats may, in the case of need, have access 
to some information that did not seem to be of any relevance at the time 
of its gathering. Those few other exceptions, when learning produces 
dysteleological changes, are always most instructive with regard to the 
mechanisms which usually achieve an improved adaptation: very often 
pathological functions reveal a great deal about the physiology of normal 

We know that an animal is encouraged by success to repeat a behavior 
and with increased vigor. We also know that repeated failures discourage 
an animal from initiating a behavior. But how does an animal know what 
is a success and what is a failure? 

The three-link chain comprising appetence, the IRM, and the consum- 
matory act can be found in many organisms as an entirely closed pro- 
gram devoid of any adaptive modifiability through learning. Most prob- 
ably this was the original State of affairs when the evolution of the 
learning processes here under discussion began. Closed programs con- 
sisting of these three processes are frequently found in lower animáis, 
particularly when an action is performed but once during the course of 
an individual animal's life. Modifiability of this system, on the other 
hand, is observed in rather more highly evolved organisms, so that mod¬ 
ifiability can be regarded as an additional evolutionáry achievement. But 
to the non-modifiable, purely linear arrangement of the three linked 
mechanisms, a fourth element is added. Its function is not to report on 
success or failure; its only message is that the consummatory act has been 
carried out and that this terminates the activity for the time being. It has 
already been mentioned in Two/I/6 that the end of an instinctive activ¬ 
ity is but rarely brought about by the exhaustion of its ASP; the end of 


IV. Learning Effected by the Consequences of Behavior 

an instinctive activity is effected by a mechanism reporting the consum- 
mation of the act. As Frank Beach (1942) has shown for the male chim- 
panzee, the emptying of the seminal vesicle is reported by propriocep- 
tors, whereupon copulatory activity is terminated. 

The evolution of a regulating cycle in which the teleonomic success of 
an activity is fed back into the systems of initial behavior can hardly be 
visualized without assuming the pre-existence of a linear system com- 
posed of an initial appetitive behavior, an IRM, and an ending in a con- 
summatory action, or, instead, in a specific State of quiescence (Two/I/ 
10), or of positive affect, or reduction of a State of negative affect. Even 
on a purely theoretical, cybernetic basis, it is impossible to construct a 
model (of a regulatory cycle) that does not contain these three elements 
and, at the same time, simulates adaptive modification of the system's 
function by the influence of success or failure. Through observations and 
by means of experiments we can fully confirm this conclusión: We do not 
knozv any behavior system modifiable by success or failure that does not inelude 
these three mechanisms. 

Certainly the animal must be supplied with some information con- 
cerning the effect the activity it has just performed has had on the exter- 
nal environment. It is clear that this information can only come from the 
external environment. Our built-in schoolmaster must, therefore, possess 
pertinent knowledge enabling him to recognize the outward signs that 
indicate success or failure. Without this knowledge he would not, in the 
case of success, be able to pat us on the back and say, "Do that again," or, 
in the case of failure, scold us. In order to be able to do this, he must 
possess quite a lot of phylogenetically acquired and genetically coded 
information. In fact, he must possess something very much like an IRM, 
a "detector," as I. P. Pavlov called it, which responds selectively to 
exteroceptor and proprioceptor reports signaling success or failure. 

I am, admittedly, not a well-read man, and least of all am I well read 
in the literature of modern psychology. I will probably, therefore, give 
the wrong impression when I say that, to my limited knowledge, P. K. 
Anokhin has been the only behaviorist theorist of learning to ask the 
question, "What minimum complication must be assumed in a biocyber- 
netic thought model constructed to simúlate learning by success or fail¬ 
ure?" Ñor, as has already been pointed out, have behavior theorists ever 
raised the problem whether there might be more than one possible way 
of constructing a system which achieves this function. The reasons why 
both the behaviorists and the older ethologists neglected this problem 
have already been delineated in the Introductory History of this book. 
The first reason is that behaviorists persist in confusing teleonomy with 
teleology and refusing to have anything to do with either. The second 
reason is that they hope to be able to abstract, by statistical means, gen- 
erally valid laws prevailing in all learning processes—if not in all behav¬ 
ior. In this way they hope to find a short cut to an understanding of 
animal and human behavior without going to the trouble of analyzing 

3. Conditioned Appetitive Behavior 


the immensely complicated physiological machinery whose function is 

From the viewpoint of the cyberneticist, there seems to be little hope 
for gaining any insight into the functioning of a complicated, informa- 
tion-exploiting system consisting of a great number of very different 
parts of subsystems by controlling the input and by calculating the sta- 
tistical probability of a particular output. H. Mittelstaedt has jocularly 
compared this to a procedure of trying to understand the function of an 
automatic dispenser of railway tickets by studying the coins put into its 
slot and the tickets coming out below. This may be a misrepresentation 
of the goal at which the scientists, thus satiricized, are aiming; yet they 
seem not to be trying to analyze the mechanism ñor are they looking 
into the device. After all, in trying to concoct a flow diagram, the cyber¬ 
neticist does the same thing—he alters the input, alters the feedback, and 
he establishes input-output rules under these varius conditions. The dif- 
ference betweeen the procedure of the psychologist and that of the 
cyberneticist is only quantitative. The cyberneticist chooses a smaller 
subsystem for his study and has, thereby, in my opinión, much better 
chances for success. Provided one has some notion about the subsystems 
comprising a complicated system, and also some notion about the partic¬ 
ular function performed by each of these subsystems, then an altogether 
different procedure is indicated. In One/II/2 it was explained that, in 
any attempt to describe a system, mere common sense dictates a method 
strictly analogous to that which cyberneticists cali making a flow diagram: 
each subsystem, whose function is tolerably clear, is represented by a 
"black box" and the influence exerted by each of them on the next one 
is symbolized by an arrow, a positive and a negative sign indicating 
additive or subtractive influences. 

B. Hassenstein has constructed flow diagrams of the different types of 
learning processes and has classified them according to these cybernetic 
models (1965, 1966). For the "black boxes" which symbolize functions he 
has chosen the concepts and terms that have been developed by the eth- 
ological approach. Our assumption that oíd ethological concepts corre- 
spond to very real physiological mechanisms is strongly confirmed by 
the unconstrained manner in which, without any mutual contradiction, 
they fit into his diagrams. Furthermore, the practical applicability of Has- 
senstein's theories in the education of children and even in the training 
of dogs and horses is an additional confirmation. 

3. Conditioned Appetitive Behavior 

As has already been explained, there are many learning processes which 
the animal achieves by actively doing something, but the result is not 
the selection of an action pattern but the selection of a stimulus situation. 


IV. Learning Effected by the Consequences of Behavior 

In the particular type of conditioning I now propose to discuss, the 
selected (or conditioned) stimulus does not directly release the action by 
which it has been conditioned; instead, it releases an appetitive behavior 
directed toward that action. As has already been explained, the regulat- 
ing cycle on which all learning by success or failure is based acquires 
information by "taking cognizance" of a rewarding or a punishing expe- 
rience. As has also been explained earlier, in all regulating cycles of 
learning by means of the experienced consequences of initiated activi- 
ties, the information acquired by a rewarding or a punishing experience 
is fed back to the mechanisms of precedent behavior. These mechanisms 
can be those of the action itself or, more commonly, those of the appe¬ 
titive behavior directed at an IRM which has an unconditioned releasing 
effect. In the latter case we speak of conditioned appetitive behavior. 

An impressive example has been published by Karl von Frisch in his 
short paper, "Ein Zwergwels, der kommt wenn man ihm pfeift "—"A 
Catfish That Comes If One Whistles For It" (1923). To investigate the 
auditory sensibility of this fish, Frisch whistled before giving the fish 
some food. At first the fish gave no reaction at all to the whistle, but after 
five triáis made on five successive days the catfish responded to the whis¬ 
tle promptly by coming out of its little cave and by beginning to search 
for food. The remarkable thing is this: during the conditioning experi- 
ments, the food had been gingerly brought into contact with the barbs 
of the fish while it was lying quiescent. To this stimulation the fish had 
responded instantly by snapping and swallowing, in other words, with 
the consummatory acts of feeding behavior. What had been associated 
with the conditioned stimulus, however, was not the action elicited by 
the subsequent unconditioned stimulus. In other words, the direct elici- 
tation of the IRM following the stimulus-to-be-conditioned led to an 
association with a behavior mechanism whose function normally precedes 
that of the IRM. It is not at all rare that the feedback to preceding behav¬ 
ior mechanisms thus skips one link, that is, the performance of the con¬ 
summatory act, and the affects the next-to-last, that of appetitive 

Another example of conditioned appetitive behavior has also been 
supplied by von Frisch (1965). A honeybee which has flown repeatedly 
to a blue flower and found no néctar but has found néctar when visiting 
a yellow flower will henceforth fly to yellow flowers only, although orig- 
inally yellow as a color had no releasing effect. In other words, the insect 
searches for the situation which has proved rewarding. A simple exper- 
iment proves that it is indeed the bee's appetitive behavior which has 
become conditioned to the yellow color. An inexperienced bee is pre- 
sented with a translucent food tray standing on yellow paper. The few 
seconds which elapse between the bee's alighting and its beginning to 
suck are utilized to exchange the yellow paper for a sheet of blue paper, 
so that sucking up the sugar solution, indubitably the consummatory act 

3. Conditioned Appetitive Behavior 


of the behavior sequence, is carried out while on the blue background. 
Nevertheless, the bee learns to fly to food trays with a yellow back¬ 
ground in spite of having fed exclusively on trays with blue back- 
grounds. As Hassenstein points out, this process permits the conclusión 
that the organism has the capacity to store information for a span of time 
sufficient to associate a precedent sensory input with a subsequent 
rewarding stimulus situation that does not come into effect until some 
seconds later. 

In his now famous experiments, I. P. Pavlov achieved a quantification of 
the conditioned salivating "reflex" of dogs by inserting a cannula into 
the salivary duct and counting the drops coming out of it. The ringing 
of a bell, originally non-releasing, becomes a conditioned stimulus after 
having been presented a few times shortly before the unconditioned 
stimulus of food was applied. Naturally this was—and still is—regarded 
as the prototype of a conditioned reflex. Nevertheless it is due to a pro¬ 
cess much more complicated than the "real" conditioned reflex as 
defined by Hassenstein and discussed in Three/III/3. It differs from the 
conditioned response of blinking, described in Three/III/4, in several 
respects: the conditioned salivation response can only be established 
while the dog is hungry; in other words, the conditioned response is 
dependent on a reinforcement which is effective only as long as the 
animal's appetitive motivation is operative. In most of Pavlov's experi¬ 
ments active appetitive behavior is made invisible by shackling the dog 
to a framework so that salivation is just about the only response which 
is not precluded. Howard Lidell has told me how he once conditioned a 
dog to salivate, using the conventional Pavlovian method, whenever a 
constantly ticking metronome was made to accelerate its beat. After 
Lidell had untied his dog, however, it ran up to the metronome at once, 
whined, wagged its tail violently and pushed against the metronome 
with its nose, salivating intensely all the while—even though the met¬ 
ronome had not changed its rhythm. What had really been conditioned 
in that dog was not a reflex at all, but a rather complex and specific Sys¬ 
tem of appetitive behavior by which a dog begs food from its master, or 
a young wolf begs food from an older pack member. As Hassenstein 
points out, all these elements of behavior could not have been learned in 
the experimental situation because the fetters made that impossible. This 
example serves to clarify the difference between the true conditioned 
reflex and the conditioned appetitive behavior: in the first, the condi¬ 
tioned stimulus constantly releases the same motor pattern, as did the 
unconditioned stimulus, the differences which occur being only the 
strength in the response and the temporal sequence (ASP). In condi¬ 
tioned appetence, however, the conditioned stimulus elicits appetitive 
behavior with all its phyletically programmed variability (Two/I/10), 
independently of whether or not it has occurred during the learning 


IV. Learning Effected by the Consequences of Behavior 

Obviously, this type of learning, as well as its biological counterpart, 
the conditioning of aversions, plays a very important role in the normal 
life of animáis and is correspondingly common in the wild. At the begin- 
ning of any sort of spontaneous appetitive behavior, the animal acts in 
a way programmed to make more probable the encountering of the 
appeted stimulus situation (Two/I/10). The second phase consists pri- 
marily of the oriented approach to an object or a stimulus configuration 
by which the final act of the chain, the consummatory act, is released. 
The information about whether the environmental situation for its per¬ 
formance has been the "correct" one or not is, in many cases, received 
only during the course of this final act. In the case of failure, it is obvi¬ 
ously impossible to change anything in the machinery of the well- 
proved but rigid phylogenetically programmed consummatory activity. 
What evidently should be adaptively modified is the appetitive behavior 
which, in the present case, landed the animal in a most disappointing 
situation. As Wallace Craig very clearly said as early as 1918, the mech- 
anisms of appetitive behavior are obviously those in which modification 
has the best chance of becoming adaptive and of directing an organism's 
behavior toward the teleonomically correct situation. Thus it has become 
the main "locus of adaptive modifiability" in the chain of physiological 

But in which other mechanisms, we must ask, should we look for the 
"built-in teacher" who "knows" what to reinforce and what to extin- 
guish? One likely place to look for sources of information is the re-affer- 
ence which the animal obtains from its own consummatory activities; 
these are sensory reports (the word "sensual" is derived from this fact) 
which may be derived both from proprioceptor and exteroceptor stimu- 
lation. In discussing the experiments made by I. Eibl-Eibesfeldt (One/II/ 
9), we have already encountered the fact that the afference which the 
animal obtains while performing a certain motor pattern carries the 
information about whether the situation is "satisfactory," that is, teleon- 
omic or not. In mentioning the bone-burying activity of dogs, we met 
with the same effect (One/II/5). 

Another example serving to illustrate the point even better is fur- 
nished by a very widespread and, evidently, phylogenetically very oíd 
motor pattern of nest building in birds. In some songbirds (Oscines) and 
quite particularly in those with highly differentiated nest-building activ¬ 
ities, the movements are coupled with a most selective IRM responding 
exclusively to one particular kind of nesting material. Our social weavers 
(Philetairus socius) failed to get into the right reproductive "mood," let 
alone into the mood for nest building, until we furnished them with the 
particular kind of African grass with which their nests are built in the 
wild; only later did we find "acceptable" substitutes. On the other hand, 
there are songbirds which possess practically no innate information 
about what their nesting material should be like. Inexperienced corvine 

3. Conditioned Appetitive Behavior 


birds, jackdaws (Coloeus monedula) as well as ravens (Corvus corax), begin 
their nest-building activities by carrying all sorts of objects—pieces of 
tile, pieces of iron and even of ice, the metal parts of broken light bulbs, 
and many other things that have been noted in my records. The birds 
carry these objects to their prospective nesting places and there they per- 
form—with the objects held tightly in their bilis—the motor pattern 
which I have termed the "tremble-shove" ( Zitterschieben in Germán). 

When performing this fixed motor pattern, a bird stands in the center 
of its prospective nest and, pressing the object against the substratum, 
shoves it sideways while shaking it with a quick vibration. The bird 
holds its head slightly tilted so that an elongated object, such as a twig, 
would scrape against the ground. At first, however, no preference for 
twigs is observable in either jackdaws or ravens. It is only when they 
happen upon an object which, when treated in the way described, 
catches and gets stuck in some way, that the motor pattern increases its 
intensity. The object is then shoved and vibrated more violently; it is 
drawn back and shoved forward again in quick succession; when it 
remains really stuck, the action reaches a truly orgastic climax, after 
which there is a critical drop in intensity and, following that, a refractory 
period during which the bird has no interest in any kind of potential 
nesting material. We know quite a number of fixed motor patterns which 
end in the same manner, in an orgastic climax followed by a refractory 
period, and we know that these climaxes are usually the goal of a partic- 
ularly intense appetite. 

The point is that the information about the teleonomically "best" nest¬ 
ing material is contained in the motor pattern itself : the one and only con- 
summatory situation which affords a máximum of satisfaction is reached 
when the bird tremble-shoves elongated objects with little projections 
that make them stick to the substrate as well as to one another. The prob- 
ability, indeed, is very great that the bird, while trying to tremble-shove 
various objects at its nest site, will happen upon twigs with little, broken- 
off branches, and thus find the appropriate object for nest building. 

The built-in teacher, checking on the exteroceptor and proprioceptor 
input coming in as re-afference of a fixed motor pattern, is a physiolog- 
ical mechanism in many ways comparable to an IRM. Like an IRM, it can 
be "taken in" by stimulus situations simulating the one for which it has 
been phyletically programmed. As has been described in Two/II/1, 2, 
IRMs are prone to dystelic functions if confronted with a "supernormal" 
stimulus situation. An analogous biological "error" can be made by the 
innate teaching mechanism of the tremble-shove. In the vicinity of 
human habitations pieces of soft metal wire are often left lying about 
and if a bird, by sheer bad luck, happens on some of these while driven 
by the motivation of tremble-shoving, it will find therein a superlatively 
rewarding object: nothing else gets stuck so well and so easily, and 
affords such satisfaction; the bird will henceforth refuse to use any other 


IV. Learning Eífected by the Consequences of Behavior 

kind of building material. Otto Koenig possesses a collection of nests 
built by birds of various species, and all these nests have been made 
entirely of soft metal wire (1962). It would be worthwhile to replícate 
this unteleonomic learning process in captive birds and to test whether 
the addiction to a superlatively rewarding but biologically disastrous 
object can be corrected. I have used the word addiction intentionally and 
without quotation marks because this type of deception, the misguiding 
of a learning process, is exactly analogous to the formation of a vice. 

4. Conditioned Aversión 

In the same way as the conditioned appetence is functionally similar to 
the conditioned reflex, so the learning process now to be discussed shows 
some functional parallels to conditioned avoidance responses. In both 
cases, the difference lies in the nervous pathway taken by the learning 
process: In the classical conditioned reflex as well as in the conditioned 
avoidance response, a direct connection is formed between the condi¬ 
tioned stimulus and the response, while in the conditioned inhibition 
and in the conditioned aversión, which I intend to discuss in this section, 
adaptive information is derived from the feedback reporting success or 
failure of the animal's action. In both cases what is learned is the selec- 
tion of a stimulus situation and not that of an action pattern. 

As everyone knows, an appetite for a certain kind of food can be con¬ 
verted into a long-lasting and even permanent aversión if the eating of 
that food has ever caused severe suffering. This phenomenon is so unre- 
markable that nobody gave it any second thoughts until John Garcia's 
attention was drawn to it by a story told to him by his mother: As a child, 
she was given a big bar of chocolate just before the family boarded a 
ferryboat that made a Crossing on a very rough sea; she became very sea- 
sick and vomited; the consequence was that she retained a lasting horror 
of any sort of chocolate, yet she clearly remembered having loved choc¬ 
olate prior to her disagreeable experience. This story served as the basis 
for a series of very important experiments made by her son. The choco¬ 
late had not caused the seasickness. The conditioned stimulus associated 
with nausea, the chocolate, was clearly distinct from the unconditioned 
stimulus, the heaving of the ferryboat. But one important fact caught J. 
Garcia's attention; both the conditioned and the unconditioned stimulus 
had one criterion in common: both were able to cause intestinal trouble. 
An obvious teleonomy could be seen in this affinity of stimuli. García 
then tried to effect a similar association between stimuli concerned with 
the intake of food and unconditioned stimulus situations not involving 
intestinal processes. In rats he attempted to create conditioned aversions 
to certain types of food by applying the worst possible kinds of punish- 
ing stimuli (electric shocks, swimming in coid water, and so forth) 

4. Conditioned Aversión 


immediately after the food had been ingested. This did not influence in 
the least a rat's choice of foodstuff. If, on the other hand, he exposed the 
rat to an unconditioned stimulus producing nausea, such as a small dose 
of apomorphine or of x-rays, even a very mild dosage produced a lasting 
avoidance of the kind of food taken a short time before. By presenting 
his rats with a menú of different dishes at intervals, and by allowing the 
nauseating stimulus to impinge after one of them had been eaten. Garda 
was able to show that Wundt's oíd laws of contiguity and succedaneity 
were perfectly valid in the learning process investigated, except for one 
interesting fact: if he inserted a new dish into the accustomed sequence of 
well-known dishes, the rat always assodated the nausea with the hith- 
erto unknown food irrespective of when, in the sequence, it was pre- 
sented. The teleonomy of this particular learning mechanism is obvious. 

Evidently conditioned responses to gustatory and olfactory stimulus 
situations can be conditioned only by association with proprioceptor 
reports coming from the vegetative system and from the intestinal tract. 
It is difficult to imagine a natural situation in which gustatory stimuli 
could bode ill with respect to non-intestinal consequences.* 

These examples suffice to demónstrate the principie of conditioned 
aversión. If the perception of a neutral or even of an appetite-inspiring 
stimulus situation has been followed once or several times by a punish- 
ing experience, it becomes associated with a response of avoidance which 
can take the form of escape or of inhibition to approach. B. Hassenstein 
gives the following biocybernetic analysis with a flow diagram (Figure 

The first question concerns the variables of input and output in the learning 
process that conditions an aversión. Input variables can be regarded as fol- 
lows: first, the originally neutral and afterwards conditioned stimulus; sec- 
ond, the reports concerning punishing experiences, which, as the examples 
show, can be of a widely varying nature—fright, pain, isolation from a social 
group, and vegetative disturbances. Output variables are, at first, responses 
of avoidance or escape followed by the conditioned inhibition of approach 
to an originally "appeted" stimulus situation. 

Input-Output Relationship : a mathematical formulation is unnecessary 
because of the similarity to a formulation for learning processes. The 
response comprises avoidance, escape, and the inhibition of other behavior. 

At first it is directly released by the bad experience; then it is transferred to 
the conditioned stimulus after it has been followed once or several times by 

*Polemics should not be included in a textbook, but it must be mentioned here that J. 
Garcia had difficulties getting his results published—just as the observations by H. Lidell 
were suppressed—because they clearly demonstrated that some learning processes are 
phylogenetically programmed. This was and is unacceptable to a certain type of ideology, 
and this same ideology motivated some scientists also to attempt, by unfair means, to 
suppress sociobiology. 


IV. Learning Effected by the Consequences of Behavior 

Figure 32. Idealized and simplified functional diagram of conditioned aversión. 
The additional bent arrow going upward at the lower left is meant to indícate 
that the negative conditioning factor need not be an external stimulus but may 
be a purely vegetative State, for instance, nausea. To inelude a differentiating link 
for conditioning by punishment, as has been done for conditioning by reward, 
would have no functional meaning. The conditioned stimulus activating aversión 
can, at the same time, be the former unconditioned stimulus releasing appetitive 
behavior belonging to another system. This is indicated by the broken line. The 
internal readiness for this other behavior is not represented. (From Hassenstein, 
Verhaltensbiologie des Kindes.) 

punishment. If an unconditioned appetitive behavior existed originally, this 
is suppressed by the process. 

Flow diagram : in Figure 33, the conduction of signáis and the evaluation of 
data are symbolized by a diagram which is able to represent the stimulus- 
response relationships just described. On its right side, a subsystem of mutual 
inhibition is represented; by virtue of their interconnections, the relationship 
is unstable in the sense that, of the two signáis coming from the left side, 
the weaker one is completely suppressed while the stronger one passes 
undiminished. (Translated from the Germán) 

Conditioned aversions play a most important biological role in the 
choosing of food, particularly in "eurytrophic" animáis, those which 
feed on many different kinds of nutrients. The pertinent experiments 
made by Curt Richter were mentioned in One/II/9. In the wild, condi¬ 
tioning of avoidance indubitably occurs much more frequently than 
learning by operant conditioning, even when one ineludes in the count 
the behavior of the most intelligent and most exploratory animáis. 

5. Conditioned Action 


5. Conditioned Action 

The conditioning of an action is something other than operant condi- 
tioning, not only from the point of view taken by the cyberneticist, but 
equally so according to behaviorists' own definition. In operant condi¬ 
tioning (conditioning of the type R) learning selects, among a number of 
possible types of behavior, a particular one that, because it affords a 
reward, is retained while the others, having been tried in vain, are dis- 
carded. "Behavior selection" characterizes the operant, or stimulus R 
type of conditioning. What Hassenstein defines as conditioned action is 
the very opposite of this: a certain action that is a behavior pattern is 
chosen and immediately rewarded by a reinforcing stimulus situation 
after the animal has happened to perform it. The reward may stem from 
an altogether different behavior system, and nonetheless a functional 
connection is established between the motor behavior and whatever 
other motivation it is that has been satisfied by the reward. An extreme 
example of such a functional connection is the association between a 
motor pattern for defense against predation becoming associated with a 
piece of sugar, as in the example of Lipizzaner stallions performing the 
capriole motivated by their appetite for sugar. The result of this learning 
process, Hassenstein says, ". . . consists of a preference, on the part of the 
animal, to perform the 'rewarded' action whenever that other motivation 
is awakened. The conditioned action can be regarded as an appetitive 
behavior which, on the base of experience, has gained an additional new 
way of attaining satisfaction." The use of the word "action"—instead of 
reaction—is meant to indicate that the learning process concerns the car- 
rying out of one particular motor pattern. 

An almost unlimited number of behaviors, fixed motor patterns and 
their combinations, as well as any fortuitous sequence of movements, can 
be conditioned by the process under discussion. Many motor sequences 
are taught to circus animáis by artificially producing a certain sequence 
and rewarding this immediately afterwards. Working with the Swiss Cir¬ 
cus Knie, and with the kind help of the circus director, the processes of 
conditioning actions has been thoroughly analyzed by E. M. Dolderer 

The important criterion of the learning process is that the reward is 
derived from a behavior system entirely different from the one to which the 
conditioned action originally belongs. As mentioned above, a horse 
learning to perform the capriole, on command, furnishes a striking 
example of a conditioned action. The movement begins as the horse first 
rears on its hind legs, then jumps high into the air and, at the culmina- 
tion of the jump, kicks out with both hind legs. Pictures showing the 
horse high in the air with both hind legs stretched straight behind are 
apt to evoke, in the uninitiated, the impression that the horse is rushing 


IV. Learning Effected by the Consequences of Behavior 

forward. Therefore it must be mentioned that the animal comes down in 
exactly the same place from which it jumped upwards. The movement 
sequence can be repeated in swift succession while the horse wheels 
around a vertical axis so that the kicking is done in different directions. 
K. Zeeb has observed that, since the capriole is performed by completely 
untrained, semi-feral horses of the Duke of Croy in the park at Dülmen, 
it can be assumed that this is an innate fixed motor pattern of the horse 
(1964). Without any doubt the action serves to defend the horse against 
large carnivores; it is easy to visualize the effect of such kicking on a pack 
of wolves attacking a horse from all sides. Colonel Podhajsky, a former 
director of the Spanish Riding School in Budapest, told me in a personal 
communication in 1946 that the capriole, as a conditioned action, was 
intentionally used by armed knights in combat. 

When conditioning the capriole, the horse is first taught to rear on its 
hind legs. Subsequently, and while the command to rear is continuously 
repeated, the horse's hocks and lower hindquarters are teased with a 
whip until the horse "loses patience" and kicks its heels. To do this it has 
to jump up high, because its forelegs are already off the ground. This 
"putting through" of a series of movements would hardly be effective 
unless the whole series were programmed as a single fixed motor pattern, 
which is indeed the case, as has been proved by Zeeb's observations. 

It remains a marvelous achievement of learning that the horse is able 
to associate a motor pattern of self-defense, that is originally released by 
an immensely stressful stimulus situation, with a reward consisting of a 
caress or, at the utmost, a cube of sugar. But the power of association 
reaches even further. At the Spanish Riding School in Vienna, it was 
customary to endose the tails of the Lipizzaner stallions in nets similar 
to those used by Sikhs for their beards, in order to prevent the kicking 
heels from tearing out the beautiful long tail hair. To an observer familiar 
with the motor pattern of the capriole, it was obvious that, even while 
the attendants were pulling the nets over the tails of the stallions, the 
"mood" for performing the capriole was already beginning to surge up 
in those noble beasts. 

Another perfect example of a conditioned action was observed by Karl 
von Frisch in a Blumenau parakeet (Brotogerys tirica Gmelin), which he 
kept in a cage in his sitting room. From time to time the bird was allowed 
the freedom of the room in order to exercise. For obvious reasons, von 
Frisch chose for the bird's "constitutional" the moment immediately 
after it had dropped a dropping. The bird soon grasped the connection 
between this act and the gaining of temporary liberty, and when von 
Frisch approached the cage and requested the bird to "mach ein Batzi," 
the bird did its utmost to obey, often with only minimal and purely 
"symbolic" success. The scope and the power of association could hardly 
be demonstrated more impressively than by the harnessing of defeca¬ 
ron—of all things!—to an appetitive behavior striving for free locomo- 

5. Conditioned Action 


tion. (Swift locomotion, on the other hand, is frequently motivated by 
the urgent need to defecate.) Incidentally, this bird generalized the 
"hypothesis" that defecation merited a reward and used it also outside 
his cage, in a number of situations, to beg for food—or merely for 

If it has been said above that a great number of activities can be con¬ 
verted, by swift rewards, into conditioned actions, it seems necessary to 
emphasize here that there are also many that, even with unlimited 
patience and persistent endeavor, refuse to become conditioned to any 
kind of stimulation. What has been said in Three/III/4 concerning the 
true conditioned reflex, is equally true of the conditioned action. The 
prerequisite for all conditioning is, as postulated by Hassenstein, the pre- 
existence of a mechanism that is sensitive to the contiguity and succe- 
daneity of stimuli (Three/III/4). Many authors underrate the number of 
existent unconditionable actions and reactions; even Foppa erroneously 
ineludes the tendón reflex in his list of conditionable responses. The neu- 
rological "tendón reflex," elicited by striking the patellar tendón or ankle 
or some other tendón with a hammer when the limb is in a certain posi- 
tion, is an artificial reflex such as one induced by electrical shock. So it 
would be trivial to say that one cannot get the same knee jerk to a con¬ 
ditioned stimulus. However, it is merely an exaggeration, by strong and 
synchronous stimulation, of the normal stretch reflex. The stretch reflex 
occurs in all degrees of strength and is highly labile, depending on many 
conditions including what one is thinking at the moment and what one 
is doing with his hands. It is doubtless quite conditionable by a number 
of circumstances, but certainly not by a reward pertaining to an alto- 
gether diíferent behavior system. 

The number of actions impossible to condition is likely to be even 
higher than the number of unconditionable reactions. Sexual action pat- 
terns, for instance, refuse to become conditioned to any kind of stimuli 
pertaining to other behavior systems—at least in non-human animáis. It 
is impossible to teach a male pigeon, through food rewards, to utter his 
courtship coo, ñor can female rats be conditioned through food or water 
rewards to assume the copulation posture, no matter how hungry and 
thirsty they may be. (No rat prostitutes are known to Science.) W. Ver- 
planck tried to condition a very common comfort activity of mallards, a 
sideways shaking of the bilí, by rewarding a bird with a small piece of 
bread every time it performed this movement. A fluffing of feathers reg- 
ularly precedes the bilí shaking which is, therefore, so predictable that 
it was easy to present the reward immediately after the movement had 
taken place. The experimenter's gestures of preparation for throwing the 
reward acted as a signal that the ducks were quick to understand. They 
performed the association all right, but what they could not do was pro¬ 
duce the motor pattern. Whenever Verplanck approached the pond and 
got ready to throw the pieces of bread, the ducks performed queer 


IV. Learning Effected by the Consequences of Behavior 

uncoordinated sideways contortions of the neck, somewhat reminiscent 
of convulsions. That was as near as they could get to the bill-shake 
through performing what may be regarded as voluntary movements. To 
my regret, Verplanck never published this very interesting finding. 

Phenomena such as these raise the question of why some actions can 
be conditioned and some cannot. The fundamental prerequisite postu- 
lated by Hassenstein, e.g. the locus which is sensitive to temporal coin- 
cidence of stimuli (Three/III/4), forms a potential bridge which seems to 
exist more frequently between behavior systems that are rather closely 
allied to one another. The system of sexual activities is built rather firmly 
into the exclusive phyletic program of its own appetites, IRM's, and 
motor patterns. It is too self-sufficient to be receptive to environmental 
situations that represent unconditioned stimuli in other systems. Simi- 
larly, conditioned aversions against a certain kind of food (Three/IV/4) 
can only be formed to conditioned stimuli which stem from the intes¬ 
tinal tract and/or the vegetative nervous system. The chances of a con¬ 
ditioned action achieving a teleonomic function become obviously 
greater the closer is the normal interaction between the two systems to 
which the to-be-conditioned action and the conditioned stimulus belong. 

Also, the possibility of forming conditioned actions is obviously cor- 
related with the general evolutionary level of the organism—possibly to 
the mere size of its central nervous system, in other words, to the number 
of neurons available. It is not just a coincidence that the most striking 
examples of conditioned actions are furnished by higher mammals, Karl 
von Frisch's Blumenau parakeet being a surprising exception. There 
must be a vast number of freely available synapses if a horse is able to 
associate a predator-defense activity with patting and sugar! 

Lastly, it seems that the formation of conditioned actions is made easier 
if the activity in question is a common, constantly available motor pat- 
tern, in other words, one with an abundant endogenous impulse pro- 
duction. All the multipurpose activities discussed in Two/I/12 are open 
to associations with impulses coming from other systems. Patterns of 
locomotion, in particular, are programmed in a manner which makes it 
easy to combine them into chains of conditioned actions. As will be dis¬ 
cussed in Three/V/1, this is the origin and the most primitive form of 
motor learning. The biological importance of conditioned actions 
increases with that of motor learning and reaches its máximum level in 

6. Conditioned Appetitive Behavior Directed at Quiescence 

The learning processes described in Section 4 of this chapter comprise 
the association of a stimulus situation, which was originally neutral or 
even constituted the goal of appetitive behavior, with an inhibition or 

6. Conditioned Appetitive Behavior Directed at Quiescence 


even an avoidance. In the learning process now to be discussed, an 
unconditioned disturbing stimulus, the "primary annoyer" itself, moti- 
vates the most complicated and variegated kinds of learning. Both types, 
conditioning of the type S as well as of the type R, can occur. 

In Two/I/10, purposive activities aimed at the avoidance of certain 
stimulus situations have been dealt with. As Monika Meyer-Holzapfel 
has suggested (1940), these can also be regarded as a kind of appetitive 
behavior striving for quiescence. In that section it was also explained why 
this term is particularly appropriate for all those activities that an organ- 
ism initiates to attain certain óptima of environmental conditions, such as 
those associated with humidity, temperature, illumination, and so on. 
Even internal "óptima"—or valúes of reference—are kept constant by 
appetitive behavior; hunger and thirst are certainly disturbing stimuli 
since they do not permit the animal to rest. 

That "state of quiescence" which the animal achieves by ridding itself 
of a disturbing stimulus must not be conceived of as anything like tor- 
pidity or sleep: the term is intended to indicate that the animal is deliv- 
ered from a disturbing stimulus and exempt from the appetitive behavior 
evoked by this "annoyer." In other words, the animal is now free to do 
something else. A crustacean finding the environment "too" dry, a noc¬ 
turnal rodent finding the illumination "too" strong, are both relentlessly 
driven to rid themselves of those disturbing stimuli and are thus pre- 
vented from doing anything else. Elimination of the annoyer allows the 
organism not to fall asleep, but to take up its everyday activities where 
they were interrupted by the annoyance. Habitat selection is almost 
exclusively accomplished through appetitive behavior directed at 

Almost all disturbing stimuli have one property in common: if their 
influence continúes for any length of time, they can damage the organ¬ 
ism to the extent of threatening its very existence. Averting this danger 
is, thus, of vital importance, and it is therefore easy to understand why 
the happy deliverance from a strongly disturbing stimulus situation acts 
as the most effective reinforcement known to Science, more effective, in 
fact, than the strongest reward received through any consummatory 
action. Devouring the most tasty prey or copulating with the most attrac- 
tive male or female are obviously less urgent than an escape from envi¬ 
ronmental conditions the continuation of which would endanger life. 

Clark L. Hull developed a theory according to which the relief of tensión 
is the most important reward or reinforcing factor of all learning (1943). 
This theory certainly holds true for all the processes of habitat selection 
mentioned above. There is no doubt that innumerable other learning 
mechanisms exist which conform exactly to Hull's theory, but there are 
just as many that do not. It seems somewhat forced to assume that the 
courtship and copulatory behavior of so many male animáis is motivated 
by the disturbing—and that means annoying—stimulus emanating from 


IV. Learning Effected by the Consequences of Behavior 

a distended seminal vesicle, and that the reinforcing reward lies in get- 
ting rid of that tensión. Still, the term Detumeszenztrieb (detumescence 
drive) has been coined on the basis of this assumption. 

The existence of many other kinds of reinforcement does not detract 
from the importance of HulTs theory. In particular, it is important for an 
understanding of the motivation of human behavior because man, more 
than any other known species, is harassed by stress stemming from the 
multiplicity of his interests and the enormous probability of conflict 
among them. So man, and modern man more than any of his precursors, 
stands in need of relief from tensión. Besieged by armies of annoying 
conflicts, he is exceedingly eager to welcome any relief of tensión and— 
unfortunately—can find this all too easily in the pharmaceutical influ- 
ence of drugs. The evolutionary construction of the human nervous Sys¬ 
tem could not foresee the danger contained in the fact that a learning 
mechanism, whose teleonomic function lies in rewarding the successful 
mastering of conflicts and other stressful problems, would respond with 
equal satisfaction to the reinforcement afforded by tranquillizing drugs. 

Stress is difficult to define because what is "still" normal and what is 
"already" pathological are next to impossible to sepárate by definition. A 
certain amount of nervous stimulation demanding activity is indispens¬ 
able to the maintenance of a healthy State of general arousal. Any excess 
of stimulation, even if it remains qualitatively identical, leads to patho¬ 
logical phenomena, particularly to neuroses—or else to addiction. 

Jules Massermann (1943) succeeded in simulating human alcoholism 
in cats. He put each experimental animal under steadily increased stress 
by making a discrimination problem gradually more and more difficult, 
and finally insoluble. Simultaneously he offered each cat, besides puré 
milk, a saucer with alcohol diluted in milk. This latter was refused at 
first, even when the alcohol was given as a minimal part of the whole. 
With increasing stress, however, a cat became more and more prone to 
accept the alcoholic drink. It is understandable that, having once discov- 
ered that alcohol affords relief from tensión, they learned to resort to 
drink, but why they first tried it still remains an unsolved question. 
Relief from tensión through the effect of alcohol was obvious. This was 
even accompanied, at first, by an improvement in a cat's discriminating 
powers, though soon afterwards these declined rapidly. Unlike humans, 
the cats were cured of their addiction immediately after the stress situa- 
tion was removed. Interestingly enough, daily occupation with similar, 
but easier problems effected a more rapid cure than complete rest. This 
demonstrates that only a quantitative difference separates "healthy" 
stress from "pathogenic" stress. 

The enormous motivating power exerted by the disturbing stimulus 
explains why conditioning of the type R, except in exploratory behavior, 
occurs almost exclusively in the context of appetitive behavior directed 
at quiescence. It seems that, except in exploration, a quite exceptional 

7. Operant Conditioning 


strength of motivation is necessary to impel an animal to try one behav- 
ior pattern after the other. One is almost tempted to say, anthropomorph- 
ically, that the animal must be in a State of desperation. The dog that 
tried several means of getting his bitch (Three/IV/1), is an exception 
which actually proves the rule: he was also in an exceptional State cióse 
to despair. The classical puzzle box represents to the cat enclosed in it an 
extremely alarming situation because it unambiguously forebodes death 
should the animal not succeed in getting out. 

7. Operant Conditioning (In the Sense Here Advocated) 

The term "operant conditioning" is generally used to identify two 
entirely different processes. If a pigeon is confined to a box in which 
there are "keys" of various colors, the bird will naturally peck at many 
different objects while it is in the box. Pecking randomly, and now and 
then striking the various keys, the pigeon will finally strike the one key 
intended by the experimenter as the rewarding one. Thereupon, by way 
of a chute, he lets a food pellet slide into the box. After a few such rep- 
etitions, the bird "grasps" that this one key is rewarding while the other 
keys are not, and it will cease to peck at the latter. The pecking has not 
changed, ñor has the motivation of the pigeon; an appetite for food has 
remained the same throughout. "Pecking at a red key" is regarded as one 
behavior learned through operant conditioning; pecking at a blue key 
would be regarded as another. 

As has already been explained in the section describing conditioned 
action, many movements can be reinforced by means of a reward pre- 
sented immediately after the movement has been completed. If an exper¬ 
imenter waits until a pigeon happens to turn its head to the left, and 
then immediately drops a food pellet, the pigeon will "form the hypoth- 
esis," after surprisingly few repetitions, that turning to the left is reward¬ 
ing. The pigeon will then learn that more pronounced turns to the left 
are even more rewarding and, in this way, it becomes possible for the 
experimenter to condition the bird to perform a circular dance in the 
desired direction. According to the oíd law of succedaneity formulated 
by Wilhelm Wundt at the turn of the century, the reinforcing effect is 
the stronger the quicker it follows the performance of the behavior that 
is to be conditioned. In her book, Lads Before the Wind (1975), Karen Pryor 
relates an amusing story illustrating this effect. A young porpoise trainer, 
an ambitious but also somewhat nervous young man, repeatedly blew 
the whistle—the expression of approval that is sufficient as a reward for 
these intelligent animáis—just a fraction of a second too soon. The por¬ 
poise, most "logically," "formed the hypothesis" that it was supposed 
only almost to touch a certain key and, consequently, it indulged, as 
Karen expresses this, "in ecstasies of almostness." 


IV. Learning Effected by the Consequences of Behavior 

It is also regarded as operant conditioning when a cat, confined in a 
puzzle box, tries many different behavior patterns—wedging its head 
between bars or into corners that might promise a way out, scratching at 
the floor, running aimlessly to and fro, uttering the distress cali of its 
species, and so on—and finally (probably while performing the digging 
movements of scratching) trips the lever that unlocks the door. 

It must be realized in what an enormously high State of general exci- 
tation the animal is when enclosed in such a box. If this situation were 
to occur within a natural setting, the inability to get out of such a con- 
finement would mean certain death, and it is perfectly understandable, 
from the teleonomic viewpoint, that- in such a situation the organism 
puts all it has into a bid for freedom. One can speculate that, physiolog- 
ically, it is the very high general excitation that breaks out into sev- 
eral different pathways. This suspicion is confirmed by the occasional 
observation of senseless, "crazy" combinations of movements charac- 
teristic of the "state of despair" in which the experimental animal finds 

I would suggest that the two learning processes here described, the 
one in pigeons and porpoises, the other in cats, should be separated con- 
ceptually. The first type (which corresponds to many textbook paradigms 
of operant conditioning) cannot be regarded as type R conditioning 
insofar as it is not the behavior that is being selected. Furthermore, the 
behavior is not selected by the animal but by operations performed by 
the experimenter; the experimenter has selected the behavior as well as 
the stimulus. One might almost wish to cali this "operated condition¬ 
ing." Also, there are often extremely few behavior patterns to choose 
from; there is precious little a confined pigeon can do besides peck. No 
high, general excitation is needed and the result of this experimenter- 
operated conditioning is, in practically all cases, a gradual increase of the 

The second type of learning process certainly does conform more 
closely to the definition that ineludes selection of behavior. Not only are 
there more behavior patterns available to be selected from, but also the 
selection is done by the organism itself and not by the experimenter. 
There are additional essential aspeets of the learning process that I 
should like to desígnate as characteristic of operant conditioning. One is 
the need for a high, general excitation that is not identical with the spe- 
cific motivation of any one of the motor patterns released and which rep- 
resents the material from which one pattern is finally selected by the 
process. Another is the immediate effect of true operant conditioning: a 
dog that, after days of unsuccessful attempts, has finally, and perhaps 
only once, succeeded in literally worming its way out of a detested ken- 
nel, will immediately resort to the same method if ever confined in the 
same kennel again. 

7. Operant Conditioning 


From the cyberneticist's viewpoint (and here I am following Hassen- 
stein's conceptualizations), operant conditioning of the type first dis- 
cussed in this section should be regarded as a formation of chains of con- 
ditioned actions. This has been discussed in Three/IV/5. Such formations 
are to be found very frequently in nature, and it is my considered opin¬ 
ión that all motor learning is based on this learning process, as will be 
discussed in Chapter V. 

Operant conditioning, in the strict sense here advocated, is rare under 
natural conditions. I had difficulty finding an example of appetitive 
behavior not being used in trying to get rid of a disturbing stimulus but 
aiming at the satisfactory discharge of a consummatory act, that is, appe¬ 
titive behavior representing the motive of the behavior selection. One of 
my dogs, trying to get to a bitch in heat confined in a kennel, performed 
the following behavior selection. First he tried to jump over the kennel's 
slightly overhanging fence. Failing to get over the fence by jumping it, 
he tried to tear through the wire of the fence using his teeth, pushing 
and pulling alternately at the wire meshwork—a behavior he had 
learned in a successful attempt to get at rabbits in their hutch. Because 
the kennel wire proved to be too strong, the dog desisted, bleeding 
slightly at the mouth. After this he seemed to give up, and we also ceased 
to feel the need for any further guarding of the kennel. During the night, 
however, the dog succeeded in digging a tunnel under the wire fence, 
although this had been set quite deeply in the ground. 

The most important function that operant conditioning performs in 
nature is closely linked with that of exploratory behavior. In exploratory 
behavior it is the stimulus situation that remains constant; the same sit- 
uation or object elicits, in the exploring animal (or in man), a whole 
gamut of behaviors tried out one after the other. In this case, a new moti- 
vating agent comes into being that is unspecific with regard to the several 
behavior systems it is capable of motivating. This will be more thor- 
oughly discussed in Chapter VI. For the present it suffices to say that the 
exploring animal can perform, through exploratory behavior, action pat- 
terns of attack, of fleeing, of eating, and such like, only as long as none 
of the specific motivations for these activities is aroused; the animal can- 
not continué its exploring when it really becomes furious, or afraid, or 
hungry. Similarly, play can only take place when all "serious" motiva¬ 
tions remain silent, that is, within a "field devoid of tensión"— im ents- 
pannten Felá, as Gustav Bally has put it, using the terminology of Kurt 

It is only this kind of learning process for which I here suggest the 
term operant conditioning. Only this type of learning mechanism is ever 
functionally integrated with exploratory behavior. This integrated unit 
plays such an immensely important role in human behavior that this, 
alone, should be sufficient to justify the use of a special term. 


IV. Learning Effected by the Consequences of Behavior 

8. Chapter Summary 

1. The New Feedback A very particular process of learning is founded 
on the principie of feeding back, to the mechanisms of initial behav¬ 
ior, the final success or failure of the activity, and of adaptively mod- 
ifying the activity accordingly. If we keep strictly to the behaviorists' 
definition of the type S and the type R of conditioning, the type of 
learning here under discussion does not fit either of the two cate- 
gories. The definition of the type S, or classical conditioning, postu- 
lates that the organism remains passive and a given type of behavior 
is conditioned to a new stimulus. The definition of operant, or type R 
conditioning, demands that the animal choose, by trial and error, a 
pattern of behavior fitting the environmental situation. What actually 
happens most frequently is that the animal gains information by 
doing something, but what it does is perform, or try to perform, a 
given behavior pattern, and gain information not concerning this 
motor pattern, but concerning the stimulus configuration in which 
this same motor pattern attains a reinforcing reward. It is not the 
behavior that is selected to fit an environmental situation, but an 
environmental situation is chosen so as to afford optimal stimulation 
for a predetermined pattern of activity. Appetitive behavior can 
equally be aimed at the satisfactory accomplishment of a consumma- 
tory action (Wallace Craig 1918) or at a State of quiescence whose pre- 
requisite is the absence of a certain disturbing stimulus. Craig termed 
this latter type of behavior "aversión," M. Meyer-Holzapfel (1940) 
described it as an appetite for quiescence. 

2. The Necessary Minimal Complication of the System The prerequi- 
site for learning by success or failure is a "feedback mechanism," in 
other words, a regulating cycle. This presupposes a relatively high 
minimal complication of the system. The best method for ascertaining 
the minimal complication of a system is an attempt to construct a flow 
diagram simulating a system's functions in the manner used in bio- 
cybernetics. As Hassenstein has shown, a flow diagram representing 
the process of learning by success or failure has to inelude, in the form 
of "black boxes," the mechanisms of appetitive behavior, of IRM, and 
of consummatory action. These three links are to be found in inde- 
pendently functioning and unmodifiable systems. The pre-existence 
of these systems must be postulated as a prerequisite in order that 
learning by success or failure could evolve. 

Even in unmodifiable systems there are feedback mechanisms 
reporting the accomplishment of a consummatory act and terminating 
the activity without, however, having furnished any information 
about its teleonomic effect. To achieve this latter function, it is nec¬ 
essary to supply information about the effect which the organism's 
activity has had on the external environment, henee an exteroceptor 

8. Chapter Summary 


apparatus is needed which possesses sufficient phyletic information 
to distinguish success from failure. This apparatus is the "built-in 
teacher" rewarding or punishing the pupil. 

3. Conditioned Appetitive Behavior In a majority of cases the reward¬ 
ing effect is associated with the precedent appetitive behavior, less 
often with the consummatory act and still less often with a State of 
quiescence, representing the "goal of aversions" in Craig's meaning 
of the term. Even upon receiving a stimulus-to-be-conditioned simul- 
taneously with the releasing of a consummatory act, the animal asso- 
ciates this not with the latter, but with the appetitive behavior phy- 
logenetically programmed to precede that act. 

Conditioned appetitive behavior has been confused with the clas- 
sical conditioned reflex. In the latter, however, the conditioned stim- 
ulus releases exclusively that one response which previously was elic- 
ited by the unconditioned stimulus, while in conditioned appetitive 
behavior it releases appetitive behavior, irrespective of whether it has 
occurred within the training situation or not. 

Very often the built-in teaching mechanism is situated in the fixed 
motor pattern of the consummatory action itself; its unmodifiable 
form furnishes rewarding reafferences only in the teleonomically cor- 
rect environmental situation; in any other it acts as a punishment by 
being "disappointing." 

4. Conditioned Aversión If a situation that is primarily neutral, or even 
eliciting appetitive behavior, is followed immediately by intense pun¬ 
ishment, strong avoidance responses become associated with this sit¬ 
uation, particularly if the punishing stimuli impinge immediately 
after appetitive behavior has begun to emerge and before the organ- 
ism has had time to proceed to the subsequent consummatory act. 
Conditioned aversions play an extremely important role in the food 
selection of eurytrophous animáis. 

5. Conditioned Action Even an isolated action pattern, phyletically 
fixed or acquired, can be associated with a reinforcing stimulus 
derived from an altogether different behavior system, provided that 
the reward is presented immediately after the action. As a result of 
this, the action is dissociated completely from its original function 
and is performed when the appetitive behavior directed at the con¬ 
ditioned stimulus becomes activated. Thus a horse learns to kick its 
heels in order to get sugar as a reward; in other words, an action 
derived from a system of defense against predators is harnessed to an 
appetitive behavior aimed at getting food. 

A conditioning by reinforcement of actions stemming from an alto¬ 
gether different system of behavior seems to occur only in very highly 
evolved organisms. On the other hand, conditioning an action by 
means of a reward stemming from a related system can lead to the 
formation of chain activities, of which each can be regarded as an 


IV. Learning Effected by the Consequences of Behavior 

appetitive behavior striving for the next. In this way the most primi- 
tive forms of motor learning are accomplished. 

6. Conditioned Appetitive Behavior Directed at Quiescence As dis- 
cussed in Two/IV/1, the unconditioned avoidance of disturbing stim- 
uli, the "aversión" according to the conceptualization of W. Craig 
(1918), can also by conceptualized as an appetitive behavior aimed at 
becoming free from these "annoyers." Among all the rewards attained 
by appetitive behavior, deliverance from disturbing stimuli exerts the 
strongest reinforcing influence. Most disturbing stimuli signify a dan- 
ger to survival if their influence endures for any length of time; there- 
fore it is teleonomically understandable that the organism tries "by 
any means at its disposal" to rid itself of them; in other words, it tries 
different motor patterns, one after the other, and learns to employ the 
most successful one. Except in the course of exploratory behavior, to 
be discussed in Three/VI/7, operant conditioning under natural cir- 
cumstances occurs almost exclusively under the excessively strong 
motivation of an appetite for quiescence. 

7. Operant Conditioning The concept of operant conditioning is here 
coníined to learning processes in which not the stimulus situation but 
the behavior pattern is selected, and from other patterns among those 
in repertory of the species concerned. Operant conditioning is not 
involved when an animal, by attempting to discharge a certain behav¬ 
ior pattern in different environmental situations, learns by trial and 
error which of these environmental situations affords a máximum of 
rewarding feedback. Ñor is this trial and error behavior to be equated 
with exploratory behavior. Exploratory behavior is based on a much 
more complicated program. In nature, operant conditioning is much 
rarer than generally assumed and occurs mainly under the influence 
of disturbing stimulus situations, in other words, within a context of 
appetence directed at quiescence. Otherwise, operant conditioning 
functions mainly in cooperation with exploratory behavior. 

Chapter V 

Motor Learning, Voluntary 
Movement, and Insight 

1. Motor Learning 

All of the learning processes dealt with in Three/III/1, 2 accomplish an 
adaptive modification within the sensory sector only. Even in the mech- 
anisms discussed in Three/III/3, 4, which function on the basis of a feed- 
back of success or failure, no adaptive change is wrought in the motor 
pattern concerned. Motor learning becomes possible only by associating 
actions, as described in Three/IV/5. Simple and constantly available 
"multipurpose" motor patterns can be wrought into chains which can be 
correctly regarded as "new," that is, not phyletically programmed move- 
ments. To the best of my knowledge the Viennese zoologist, Otto Storch 
(1949), was the first to cali attention to the fact that Erwerbsrezeptorik, 
"receptor learning" preceded Erwerbsmotorik, "motor learning" in evo- 
lution, and was much more common, especially among lower animáis, 
than motor learning. 

The most primitive form of a "new" motor pattern acquired through 
learning is the linking together of a chain of conditioned actions as 
described in Three/IV/5. If a reinforcing stimulus is made to follow 
immediately upon some readily available motor pattern, especially one 
of the multipurpose type activities, and if this process is repeated several 
times in quick succession, a long sequence can be formed of the repeti- 
tion of this same motor pattern. For instance, the reinforcement of a 
slight turn to the left can teach a pigeon to repeat it as a kind of dance 
round and round in that direction. The pigeon has formed the "hypoth- 
esis" that turning to the left causes grain to fall out of a chute. 

For reasons already discussed in Three/I/12 and Three/I/5, among all 
the fixed motor patterns, those of locomotion are the most easily welded 


V. Motor Learning, Voluntary Movement, and Insight 

into long sequences of conditioned actions. As the examples that were 
cited demonstrated, most of the conditioned actions produced during the 
training of horses are derived from this source. In laboratory experiments 
it is possible, by rewarding the first movement of a desired sequence and 
by reinforcing a second one when the animal happens to produce it, and 
by persevering in this procedure, to form amazingly complicated chains 
of conditioned responses, for instance, two pigeons "playing ping- 
pong," that is, driving a celluloid ball back and forth by pecking at it. 
Such chains of conditioned actions can, of course, be regarded as new 
motor patterns acquired by learning. Very striking examples of this pro- 
cess of "shaping" skilled movements are to be found in Karen Pryor's 
book, Lads Before the Wind (1975). 

In the wild, this type of motor learning plays an important role in the 
acquisition of path habits. O. Koehler and W. Dingler demonstrated this 
process with a highly instructive film (1953). A mouse is put into a so- 
called high labyrinth, that is, a maze constructed with pieces of lath and 
elevated on stilts, high enough to discourage the animal from jumping 
down. At first, the mouse creeps along, step by step, very slowly feeling 
its way by groping about with its tactile vibrissae. After having moved 
through the maze two or three times in this painstaking way, the mouse 
begins to traverse certain stretches of the maze, at first very short ones, 
at a fast run, after which it balks again and returns to the vibrissae-pal- 
pating walk. The two methods of proceeding, first the slow walk steered 
by mechanisms of exploiting instant information, and second the fast 
run, which represents a learned or "skilled" movement, continué to 
altérnate for a long time. The stretches mastered by skilled movement 
increase in length; they appear at new places along the course and fuse 
at their ends. Very often, though, the places of their fusing remain visi¬ 
ble in the form of a slight hesitation made by the animal when passing 
those particular points—much like the way a child gets stuck again and 
again at certain places when reciting a poem or playing a piece of music. 
In the end, these remaining difficulties are smoothed over and disappear, 
so that henceforth running through the maze constitutes a single skilled 
flowing movement. 

The telenomic valué of this kind of path learning is obvious: An entire 
passage along a complicated route can be negotiated without any retarda- 
tion through reaction times. For this reason the performance is very much 
faster than any analogous negotiation based on instant information- 
exploiting mechanisms could ever be. This is one—often forgotten—rea- 
son why animáis confine themselves to a spatially limited territory: 
within it they have mastered all the possible contingencies of locomotion 
by skilled movements. The difference between the speed of skilled 
movements and those steered by instant information is brought home 
most impressively to anyone who has ever tried to catch a mouse, a liz- 
ard, a blenny, or any coral fish. Along its known pathways, the animal 

1. Motor Learning 


is much too quick to be caught, but if one can succeed in stampeding it 
out of its known territory, the odds are that one will catch it. 

In his book, Die Orientierung der Tiere im Raum (1919), Alfred Kühn 
described a form of learning which he postulated as being a possibility, 
and which he called "mnemotaxis." He assumed an engram containing 
the entire sequence of motor patterns, which during an action was syn- 
chronized with the sequence of incoming stimuli, thus confirming the 
correctness of movement at every step. This he termed mnemische Hom- 
ophonie. Critics argued that the process of orientation could only function 
if the animal, while running along a learned pathway, stepped exactly 
in the same places each time and thus received exactly the same 
"expected" reafference. In the second edition of his book, Kühn deleted 
"mnemotaxis" because no example was known of an animal behaving in 
the postulated manner. Since then, such animáis have been discovered. 
The water shrew (Neomys fodiens) seems to achieve a fusing of single 
locomotor patterns into one skilled flowing movement of running along 
habitual paths by joining the elements, through conditioning, each to 
the stimulus situation, which not only provokes the next but also orients 
the animal and tells it that it is still on the right path. If this is true, then 
the animal must become disoriented and its skilled movements broken 
up, if only one step misses an appointed place. Contrary to the expecta- 
tions of Kühn's critics, this is quite exactly what happens to the water 
shrew—and very probably to many other small mammals. If, experimen- 
tally, one causes a break in the shrew's "mnemic homophony" by chang- 
ing a bit of the ground traversed by its habitual path, it will come to a 
complete stop, begin to explore its surroundings with palpating vibris- 
sae, and run back along a stretch of the way it has come. Recognizing 
some landmark passed a few seconds before, it will reorient, resume the 
former course and, by "gathering momentum," attempt to overeóme the 
difficulty caused by the changed bit of ground. When reciting poems, 
children do exactly the same thing. In one place, just before reaching 
"home," the path habit of my water shrews involved the necessity of 
jumping up onto the lid of their wooden nest box that was several inches 
high. After I removed this box, the shrews jumped up onto emptiness 
and, after falling down, continued to repeat the movement several times. 
Curiously enough, the shrews were demonstrably able to see the box—it 
contained their nest—because, when I put them onto the floor, they ran 
to the box at once, even when it was half a meter away. Thus the skilled 
and flowing movement of path running was actually able to overeóme 
the influence of instant information! 

As I mentioned in Three/III/2, little is known about the physiology of 
association. We do not know how the motor elements pertaining to 
skilled movements are linked together. Some researchers have assumed 
that proprioception takes part in the process of learning a sequence of 
movements "by heart," such as playing a musical instrument, reciting a 


V. Motor Learning, Voluntary Movement, and Insight 

poem, or running through a maze. The term "kinesthetic" (from the 
Greek kinesis , "the movement," and aisthanomai, "I feel") has been used 
in this context. Subjectively, when we "visualize" ourselves performing 
such a skilled movement, we seem to have some "feeling" of its perfor¬ 
mance. Also, in the process of acquiring a motor skill, proprioceptor 
action may play a role. 

But two arguments oppose the kinesthetic hypothesis. As Erich von 
Holst has shown, skilled movements are subject to all the laws of relative 
coordination (Two/I/13) and of magnet effect that prevail in centrally 
coordinated movements; in centrally coordinated movements propri¬ 
oceptor processes play no part at all. Furthermore, it is known from the 
work done by J. Eccles that skilled motor patterns are formed in and con- 
trolled by the cerebellum. Self-observation, too, argües against the 
assumption that the faculty for performing skilled movements is based 
on a kinesthetic memory. If you ask someone who is so adept at driving 
an automobile that the motions of controlling the car have become "sec- 
ond nature," which foot is used on the brake and which on the clutch, 
it can be observed with predictable certainty that the person will move 
both feet alternately, in other words, the person has to activate the 
skilled movements in order to determine, by self-observation, which foot 
does what. 

It is justified to assume that skilled movements more complicated than 
those of running along a known path are formed in the same way. There 
is no reason why simultaneously activated motor elements should not be 
associated by means of the same process that links successive ones. 

Curiously enough, skilled motor patterns, particularly after they have 
been in use for some time and are well "ground in," develop some of the 
characteristics of centrally coordinated fixed motor patterns. As already 
mentioned, all the laws of central coordination prevail in skilled move¬ 
ments and tend to forcé the several motor elements into harmonious 
integral frequences. The better the possibilities for integration are, the 
more stable are the skilled movements. Conversely, the skilled move¬ 
ment is that much more unstable—and the more difficult to learn—the 
more the several frequencies of the elements resist such a harmonious 
integration. Everyone who, as a child, was taught to play the piano, will 
remember how hard it was to play triplets with one hand and quavers 
with the other. The harmonizing effects of central coordination endow 
skilled movements with forms at once economic and elegant, which have 
a strong appeal to our aesthetic sense. 

A second property shared by the skilled movement and the fixed 
motor pattern is their strong resistance to any change. Karl Bühler's 
assumption that everything learned can be wiped out and forgotten 
(1922) holds true for all learning by reinforcement; it does not, however, 
hold true for imprinting (Three/VI/3), ñor does it for the acquisition of 
skilled movements. The professional coach of sports such as tennis or 

2. So-Called Voluntary Movement 


swimming knows very well that it is hopeless to attain máximum effi- 
ciency for a person who, through untutored training, has acquired self- 
taught motor habits. These most perniciously block the way for learning 
those motor habits indispensable to optimum skill. 

The third and most amazing similarity between skilled movements 
and fixed motor patterns is that, when not used for a certain period of 
time, both give rise to appetitive behavior. As H. Harlow has shown, 
rhesus monkeys take such pleasure in performing complicated move¬ 
ments, such as opening a lock, that they not only do this independently 
of any reward, but even go to considerable trouble in order to be allowed 
to do so. Karl Bühler was the first to notice this phenomenon and he 
spoke about Funktionslust —pleasure in the function. The strength of the 
motivation generated by this pleasure is evident from the avidity for 
dancing, skiing, and other activities that we, ourselves, display. One 
example can illustrate the blind striving for the performance of skilled 
movements: if a military rifle is presented to a man who has seen military 
Service, he will predictably take it up and try to perform some learned 
actions with this objectionable object, however much he may have hated 

As we know from self-observation, the "smoothing out" which makes 
the skilled movement so elegant and economically efficient affords a cer¬ 
tain pleasure that is qualitatively unmistakable, and is the greater the 
more complicated the movement and the more difficult the accomplish- 
ment is to attain. The teleonomy of the learning process thus motivated 
is obvious, as every "smoothing," that is, removing any roughness, 
means a saving of energy. In an earlier book I wrote about a "perfection- 
reinforcing mechanism." It is my belief that this mechanism has become 
liberated from its original teleonomic function in humans, and this "run- 
ning free" has become the root of all human arts, the oldest of which is 

2. So-Called Voluntary Movement 

One definition of the word "voluntary," which is going to be applied to 
the processes now discussed, is that given in the Concise Oxford Dictionary: 
"control exercised by delibérate purpose over impulse." An inclusión of 
purposiveness in the definition of voluntary movements is admissible to 
the extent that voluntary movements, with the few exceptions men- 
tioned at the end of the last section, always form part of some sort of 
appetitive behavior. 

As was mentioned in Three/IV/5, an important physiological differ- 
ence exists between those motor patterns which most easily form con- 
ditioned actions by becoming associated with some reinforcement and 
those others which persistently refuse to do so. Examples of both have 


V. Motor Learning, Voluntary Movement, and Insight 

been given in this section. We do not know what, physiologically, causes 
this difference, but we can make some suggestions about their teleon- 
omy. Those fixed motor patterns which are easily built into skilled move- 
ments almost always belong to the "multipurpose" type of activity dis- 
cussed in Two/I/11. An obvious part of their phylogenetic program is to 
serve as elements for the formation of skilled movements. The non-con- 
ditionable motor patterns are primarily those which normally appear 
only when driven by their own particular motivation. The elements out 
of which skilled movements are composed are predominantly common, 
easily "available" patterns such as pecking in pigeons, gnawing in 
rodents, and, more than any others, the movements of locomotion, 
including those of taxis-controlled turnings in every direction. 

Regarding the patterns of locomotion, we are practically certain that 
their source is to be found in endogenous impulse production and that 
their form is determined by central coordination, as was assumed by 
Erich von Holst. This assumption implies that the motor patterns them- 
selves are not susceptible to adaptive change through environmental 
stimulation. Whenever adaptive change becomes necessary, it is effected 
by a superposition of physiologically quite different motor processes 
which are directly controlled by external stimuli. Erich von Holst sub- 
sumed these under the concept of a "mantle of reflexes" which envelops 
the hard and immutable core of centrally coordinated fixed patterns. 
When a dog is trotting over an uneven surface, taxis-controlled move¬ 
ments are, at every step, interposed adaptively between the centrally 
coordinated movement and the irregularities of the ground. 

The environment through which an animal moves often demands 
such radical and sudden acts of adaptation from an animal's locomotion 
that the "mantle of reflexes," in other words, the superposition of taxis- 
controlled movements, fails to cope with the situation. In this case, the 
central nervous system has to resort to the faculty that in Two/VI/11 was 
likened to the limited power exerted by a ship's captain who cannot 
change the movements of the ship's machinery but can command it to 
stop and to go ahead again. The phyletic adaptation of locomotor move¬ 
ments to extreme environmental conditions was not achieved, as some 
students of behavior believed, by "softening" fixed motor patterns so as 
to make them fit the requirements. Phylogeny found quite a different 
way to make an animal's locomotion adaptable to the ever-changing 
demands made on it by the circumstances of environments possessing 
complicated and irregular spatial structures, as do the crags of a rocky 
mountain or the branches of a tree. What actually happened during the 
evolution of locomotor patterns can be deduced from a comparison of 
closely related species, some of which live in spatially simple and regular 
biotopes, while others inhabit rocky habitats or trees. The more spatially 
homogeneous the biotope, the fewer are the demands made on instant 
adaptation of locomotion. The complete homogeneity of the open sea 

2. So-Called Voluntary Movement 


makes it unnecessary for some freely moving creatures, such as the jel- 
lyfish Rhizostoma pulmo, to react spatially to any sort of obstacle. Even in 
fast-swimming pelagic animáis, the spatial responses helping to avoid 
solid obstacles need not be any greater than that of the ship's captain in 
our illustration who can command only deviations involving a rather 
large radius, and Controls the engines only to the extent of moving for- 
ward, stopping and going into reverse. Among terrestrial habitats, the 
open steppe is somewhat similar to the open sea with regard to homo- 
geneity, at least in two dimensions. The locomotion of steppe-dwelling 
animáis is often surprisingly unadaptive as far as irregularities of the 
ground and unforeseen obstacles are concerned. The various gaits of 
walking, trotting, or galloping can only be "commanded" as unitary 
entities, each used at different speeds, the slower one being exchanged 
for the faster one at a predictably specified pace, much as the gears are 
changed by the automatic gear shift of a car. To a cantering horse or 
antelope, the fíat ground gives very much the same support at every 
bound or leap, and even when it fails to do so, the obstacle can usually 
be seen far enough ahead so that swerving aside or stopping in time is 
possible. The sudden appearance of obstacles causes horses and other 
steppe-dwelling animáis to stumble or fall. 

Both open sea and steppe-dwelling organisms command only a rather 
poor capacity for superimposing taxis-controlled movements on their 
locomotor patterns. A horse walking uphill over uneven ground does 
not just struggle forward blindly; it does pay a modicum of attention to 
the ground it is traversing. But it succeeds only very inaccurately in put- 
ting its hooves on those places offering firm support. 

In order to effect a more differentiated adaptation of locomotion to the 
substratum, the independently available elementary motor pattern must 
be made smaller. To aim one locomotor element exactly at a certain sup¬ 
port, the "minimum separable unit" should not be a series of stepping or 
trotting or galloping movements, but one step or jump. That element 
must be independently available for the simple reason that an animal is 
often forced to aim its step or jump at a narrow goal, for instance, the 
summit of a rock, and to stop there. It is quite surprising how widely 
creatures closely related zoologically differ with regard to this ability. 
Horses as well as steppe-dwelling antelope are very bad at it; the donkey 
exceeds the horse, and certain species of mountain zebra are past masters 
of the art. The mulé is proverbially surefooted because it has inherited 
some of the donkey's capabilities. The agility of the one antelope species 
specialized for living in rocky surroundings, the chamois of the Alps 
(Rupicapra rupicapra), excels even that of goats, which are adapted to 
mountain dwelling. Besides the faculty of disposing very freely of dis- 
jointed motor elements, these professional mountaineers possess an 
especially effective "mantle of reflexes," in other words, a very special 
ability to superimpose taxis-oriented movements on their locomotor pat- 


V. Motor Learning, Voluntary Movement, and Insight 

terns. A herd of chamois can pass over a slope covered with large and 
irregular boulders without interrupting the elegant and economic motor 
pattern of their gallop, which seems quite rhythmic and regular and in 
which, none the less, each step is aimed at a strictly determined spot. 
Only at intervals are slight "syncopes" observable; these show that, after 
all, the "captain" is sometimes forced to command a "stop," even if the 
interruption is but a very short one. 

An element of locomotion that has become independently available to 
very many mammals is, so to speak, a half step forward performed with 
one front leg. In difficult situations, such as when being shut up in a 
puzzle box, but also quite generally in situations of stress, animáis of the 
most diverse kind will "paw" with one front leg. They will do this long 
before they have learned through experience that this movement, when 
properly oriented, can open a latch. Conversely, researchers experiment- 
ing with animáis are the ones who have learned to construe problem 
situations in such a way as to make the pawing movement applicable for 
the animal. Unlearned pawing is used by horses to beg for food and by 
dogs to beg forgiveness. 

Among all possible habitats, the arboreal biotope makes the greatest 
demands on locomotion, both with respect to small independently avail¬ 
able motor elements and to those capacities necessary for precise orien- 
tation. Among all tree-climbing animáis, those with prehensile hands, 
such as chameleons, some marsupials, and many primates, are most 
dependent on the adaptability of locomotion. While a paw equipped 
with curved claws like those of a squirrel, or with adhesive pads like 
those of a tree frog, can catch hold even when hitting a support rather 
haphazardly, the prehensile hand does not afford any support at all 
unless it can cióse its grip at exactly the right place and at exactly the 
right moment. There is an interesting correlation between the functions 
of visual orientation and of clutching with prehensile hands: all mam¬ 
mals equipped with prehensile hands and able to move through 
branches in swift jumps, possess a "lemur-face," that is, eyes directed for¬ 
ward like those of humans, in other words, a visual apparatus specialized 
for stereoscopic orientation within space. It is easy to understand that 
man, in whose perceptual thought spatial visualization plays such a fun¬ 
damental role, could only have descended from ancestors possessing pre¬ 
hensile hands. 

The phylogenetic process of "cutting out" elements of general appli- 
cability from longer locomotor coordinations has very probably been the 
phyletic origin of all "voluntary movements." It must be kept in mind, 
though, that what we are accustomed to conceptualizing under this term 
is by no means the original, independently available element, but skilled 
movements built up by learning processes utilizing many such elements 
by linking them into most complicated chains and nets of conditioned 
actions. The elements themselves, programmed to be combined into 

3. Voluntary Movement and Insight 


skilled movements, are by no means basic neural elements. Each of them 
must be regarded as an autonomous, if small, fixed motor pattern that is 
a mechanism located on a very much higher level of integration than the 
elementary contractions of a muscle fiber—as represented on the lowest 
level of Tinbergen's diagram (Two/IV/3). The very minimum of a "vol¬ 
untary element" implies the coordinated contractions of antagonistic 
muscles, such as occur when bending and stretching a finger. 

Skilled movements are directed by our will only as far as their initia- 
tion is concerned. Their coordination, as we have seen, is subject to the 
workings of a rather rigid chain of associations which our will can 
change only gradually and by detours, and even then incompletely. 
What ought to be conceptualized as a voluntary movement is, in a strict 
sense, an unlearned movement, only commanded and coordinated by our 
"free will." If we try to coordinate our independently available motor 
patterns in a new way—that may be dictated by intelligence and 
insight—the resulting movements always look extremely awkward. A 
right-handed person trying to write with the left hand or anyone trying 
to make a drawing while looking at the paper and pencil in a mirror can 
be cited as examples. What we then observe in ourselves is strikingly 
reminiscent of the behavior of a mouse in an unfamiliar high maze. 

3. Voluntary Movement and Insight 

As was anticipated in Two/VI/1, a cióse connection exists between the 
evolutionary developmen